refactor(cli): auto-generate skill doc from command registry (#71)

Add user-facing WORKFLOW_STORAGE_ROOT environment variable to override
the default storage directory (~/.uncaged/workflow). The existing
UNCAGED_WORKFLOW_STORAGE_ROOT (internal/test) takes priority.

- Update storage-env.ts with priority chain: internal > user > default
- Add env var documentation to CLI help text
- Add 5 tests covering all priority/fallback scenarios

Fixes #63

packages/cli-workflow/__tests__/fixtures/live/logs/C9NMV6V2TQT81/01LIVECMPLT01DDDDDDDDDDDDG.data.jsonl

@@ -0,0 +1,4 @@
+{"name":"demo-live","hash":"C9NMV6V2TQT81","threadId":"01LIVECMPLT01DDDDDDDDDDDDG","parameters":{"prompt":"hello","options":{"maxRounds":5,"depth":0}},"timestamp":1714963400000}
+{"role":"planner","contentHash":"FF7YQ5W3S2EV6","meta":{"phase":"plan","flags":[1,2]},"refs":[],"timestamp":1714963201000}
+{"role":"coder","contentHash":"EN34XX1W4WAFJ","meta":{},"refs":[],"timestamp":1714963202000}
+{"returnCode":0,"summary":"fixture completed"}

packages/cli-workflow/__tests__/fixtures/live/logs/C9NMV6V2TQT81/01LIVECMPLT01DDDDDDDDDDDDG.info.jsonl

@@ -0,0 +1,2 @@
+{"tag":"DEBUGTAG1","content":"bundle loaded","timestamp":1714963400050}
+{"tag":"DEBUGTAG2","content":"multi\nline","timestamp":1714963400500}

packages/cli-workflow/__tests__/fixtures/live/logs/C9NMV6V2TQT81/01LIVEINFLY01DDDDDDDDDDDDG.data.jsonl

@@ -0,0 +1,2 @@
+{"name":"demo-live","hash":"C9NMV6V2TQT81","threadId":"01LIVEINFLY01DDDDDDDDDDDDG","parameters":{"prompt":"hello","options":{"maxRounds":5,"depth":0}},"timestamp":1714963200000}
+{"role":"planner","contentHash":"P6M9FHE1GSBN0","meta":{"x":1},"refs":[],"timestamp":1714963201000}

packages/cli-workflow/__tests__/fixtures/live/logs/C9NMV6V2TQT81/01LIVEOLDER01DDDDDDDDDDDDG.data.jsonl

@@ -0,0 +1,2 @@
+{"name":"demo-live-old","hash":"C9NMV6V2TQT81","threadId":"01LIVEOLDER01DDDDDDDDDDDDG","parameters":{"prompt":"old","options":{"maxRounds":5,"depth":0}},"timestamp":1714963000000}
+{"returnCode":0,"summary":"older thread"}

packages/cli-workflow/__tests__/fork-cli.test.ts

@@ -111,7 +111,7 @@ describe("cli fork", () => {
     const sourceData = join(storageRoot, "logs", hash, `${sourceId}.data.jsonl`);
     const sourceRunning = join(storageRoot, "logs", hash, `${sourceId}.running`);
     await waitUntilRunningAbsent(sourceRunning);
-    await waitUntilMinDataLines(sourceData, 4);
+    await waitUntilMinDataLines(sourceData, 5);
 
     const forked = await cmdFork(storageRoot, sourceId, "planner");
     expect(forked.ok).toBe(true);
@@ -122,22 +122,22 @@ describe("cli fork", () => {
     const newData = join(storageRoot, "logs", hash, `${newId}.data.jsonl`);
     const newRunning = join(storageRoot, "logs", hash, `${newId}.running`);
     await waitUntilRunningAbsent(newRunning);
-    await waitUntilMinDataLines(newData, 4);
+    await waitUntilMinDataLines(newData, 5);
 
     const text = await readFile(newData, "utf8");
     const lines = text
       .trim()
       .split("\n")
       .filter((l) => l !== "");
-    expect(lines.length).toBe(4);
+    expect(lines.length).toBe(5);
     const start = JSON.parse(lines[0] ?? "{}") as Record<string, unknown>;
     expect(start.threadId).toBe(newId);
     expect(start.forkFrom).toEqual({ threadId: sourceId });
 
-    const last = JSON.parse(lines[lines.length - 1] ?? "{}") as Record<string, unknown>;
-    expect(last.role).toBe("reviewer");
+    const lastRoleLine = JSON.parse(lines[lines.length - 2] ?? "{}") as Record<string, unknown>;
+    expect(lastRoleLine.role).toBe("reviewer");
     const cas = createCasStore(getGlobalCasDir(storageRoot));
-    expect(await getContentMerklePayload(cas, String(last.contentHash))).toBe("rev-1");
+    expect(await getContentMerklePayload(cas, String(lastRoleLine.contentHash))).toBe("rev-1");
   });
 
   test("fork without --from-role retries last role", async () => {
@@ -162,7 +162,7 @@ describe("cli fork", () => {
     const sourceData = join(storageRoot, "logs", hash, `${sourceId}.data.jsonl`);
     const sourceRunning = join(storageRoot, "logs", hash, `${sourceId}.running`);
     await waitUntilRunningAbsent(sourceRunning);
-    await waitUntilMinDataLines(sourceData, 4);
+    await waitUntilMinDataLines(sourceData, 5);
 
     const forked = await cmdFork(storageRoot, sourceId, null);
     expect(forked.ok).toBe(true);
@@ -173,23 +173,23 @@ describe("cli fork", () => {
     const newData = join(storageRoot, "logs", hash, `${newId}.data.jsonl`);
     const newRunning = join(storageRoot, "logs", hash, `${newId}.running`);
     await waitUntilRunningAbsent(newRunning);
-    await waitUntilMinDataLines(newData, 4);
+    await waitUntilMinDataLines(newData, 5);
 
     const text = await readFile(newData, "utf8");
     const lines = text
       .trim()
       .split("\n")
       .filter((l) => l !== "");
-    expect(lines.length).toBe(4);
+    expect(lines.length).toBe(5);
 
     const replayCoder = JSON.parse(lines[2] ?? "{}") as Record<string, unknown>;
     expect(replayCoder.role).toBe("coder");
     const cas = createCasStore(getGlobalCasDir(storageRoot));
     expect(await getContentMerklePayload(cas, String(replayCoder.contentHash))).toBe("c1");
 
-    const last = JSON.parse(lines[lines.length - 1] ?? "{}") as Record<string, unknown>;
-    expect(last.role).toBe("reviewer");
-    expect(await getContentMerklePayload(cas, String(last.contentHash))).toBe("rev-2");
+    const lastRoleLine = JSON.parse(lines[lines.length - 2] ?? "{}") as Record<string, unknown>;
+    expect(lastRoleLine.role).toBe("reviewer");
+    expect(await getContentMerklePayload(cas, String(lastRoleLine.contentHash))).toBe("rev-2");
   });
 
   test("fork rejects unknown role with available names", async () => {
@@ -213,7 +213,7 @@ describe("cli fork", () => {
     const sourceData = join(storageRoot, "logs", added.value.hash, `${sourceId}.data.jsonl`);
     const sourceRunning = join(storageRoot, "logs", added.value.hash, `${sourceId}.running`);
     await waitUntilRunningAbsent(sourceRunning);
-    await waitUntilMinDataLines(sourceData, 4);
+    await waitUntilMinDataLines(sourceData, 5);
 
     const bad = await cmdFork(storageRoot, sourceId, "ghost-role");
     expect(bad.ok).toBe(false);

packages/cli-workflow/__tests__/gc-cli.test.ts

@@ -129,7 +129,10 @@ describe("gc cli and garbageCollectCas", () => {
     });
 
     const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
-    const proc = spawnSync(process.execPath, [cliEntryPath, "gc"], { env, encoding: "utf8" });
+    const proc = spawnSync(process.execPath, [cliEntryPath, "cas", "gc"], {
+      env,
+      encoding: "utf8",
+    });
     expect(proc.status).toBe(0);
     expect(String(proc.stdout).trim()).toBe("scanned 1 threads, 2 active refs, deleted 1 entries");
   });

packages/cli-workflow/__tests__/help.test.ts

@@ -0,0 +1,84 @@
+import { describe, expect, test } from "bun:test";
+import { runCli } from "../src/cli-dispatch.js";
+import { formatSkillDoc } from "../src/cmd-help.js";
+
+const STORAGE_ROOT = "/tmp/help-test-storage";
+
+describe("help command", () => {
+  test("help returns 0", async () => {
+    const code = await runCli(STORAGE_ROOT, ["help"]);
+    expect(code).toBe(0);
+  });
+
+  test("help --skill returns 0", async () => {
+    const code = await runCli(STORAGE_ROOT, ["help", "--skill"]);
+    expect(code).toBe(0);
+  });
+});
+
+describe("formatSkillDoc", () => {
+  const doc = formatSkillDoc();
+
+  test("contains title", () => {
+    expect(doc).toContain("# uncaged-workflow CLI Reference");
+  });
+
+  test("contains all command group headers", () => {
+    expect(doc).toContain("### workflow");
+    expect(doc).toContain("### thread");
+    expect(doc).toContain("### cas");
+    expect(doc).toContain("### init");
+    expect(doc).toContain("### Top-level shortcuts");
+  });
+
+  test("contains core concepts", () => {
+    expect(doc).toContain("## Core Concepts");
+    expect(doc).toContain("Workflow");
+    expect(doc).toContain("Bundle");
+    expect(doc).toContain("Thread");
+    expect(doc).toContain("CAS");
+    expect(doc).toContain("Registry");
+  });
+
+  test("mentions all workflow subcommands", () => {
+    for (const sub of ["add", "list", "show", "rm", "history", "rollback"]) {
+      expect(doc).toContain(`workflow ${sub}`);
+    }
+  });
+
+  test("mentions all thread subcommands", () => {
+    for (const sub of [
+      "run",
+      "list",
+      "show",
+      "rm",
+      "fork",
+      "ps",
+      "kill",
+      "live",
+      "pause",
+      "resume",
+    ]) {
+      expect(doc).toContain(`thread ${sub}`);
+    }
+  });
+
+  test("mentions all cas subcommands", () => {
+    for (const sub of ["get", "put", "list", "rm", "gc"]) {
+      expect(doc).toContain(`cas ${sub}`);
+    }
+  });
+
+  test("contains exit codes section", () => {
+    expect(doc).toContain("## Exit Codes");
+  });
+
+  test("contains environment variables section", () => {
+    expect(doc).toContain("## Environment Variables");
+    expect(doc).toContain("UNCAGED_WORKFLOW_STORAGE_ROOT");
+  });
+
+  test("contains typical workflow section", () => {
+    expect(doc).toContain("## Typical Workflow");
+  });
+});

packages/cli-workflow/__tests__/live.test.ts

@@ -0,0 +1,369 @@
+import { afterEach, beforeEach, describe, expect, test } from "bun:test";
+import { spawn, spawnSync } from "node:child_process";
+import { cp, mkdir, mkdtemp, readFile, rm, writeFile } from "node:fs/promises";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { fileURLToPath } from "node:url";
+
+import { createCasStore, getGlobalCasDir, putContentMerkleNode } from "@uncaged/workflow";
+
+import {
+  formatLiveDebugLine,
+  formatLiveTimeLabel,
+  LIVE_CONTENT_MAX_LINES,
+  type LiveRoleRow,
+  renderLiveRoleStepLines,
+} from "../src/cmd-live.js";
+import { parseLiveArgv } from "../src/live-argv.js";
+
+const cliEntryPath = fileURLToPath(new URL("../src/cli.ts", import.meta.url));
+const fixtureRoot = fileURLToPath(new URL("./fixtures/live", import.meta.url));
+
+/** Bodies for Merkle content nodes; hashes must match `.data.jsonl` fixtures. */
+const LIVE_FIXTURE_PLANNER_BODY =
+  "alpha\nbeta\ngamma\nLINE4\nLINE5\nLINE6\nLINE7\nLINE8\nLINE9\nLINE10\nLINE11";
+
+describe("live helpers", () => {
+  test("formatLiveTimeLabel pads HH:MM:SS", () => {
+    const label = formatLiveTimeLabel(new Date("2024-06-01T09:08:07.000Z").getTime());
+    expect(label).toMatch(/^\d{2}:\d{2}:\d{2}$/);
+  });
+
+  test("formatLiveDebugLine flattens newlines in message", () => {
+    const line = formatLiveDebugLine(0, "TAG1", "a\nb");
+    expect(line).toContain("[TAG1]");
+    expect(line).toContain("a b");
+    expect(line).not.toContain("\n");
+  });
+
+  test("renderLiveRoleStepLines truncates content to LIVE_CONTENT_MAX_LINES", () => {
+    const lines = Array.from({ length: LIVE_CONTENT_MAX_LINES + 3 }, (_, i) => `L${i + 1}`);
+    const row: LiveRoleRow = {
+      role: "r",
+      content: lines.join("\n"),
+      meta: { k: "v" },
+      timestamp: 0,
+    };
+    const out = renderLiveRoleStepLines(row, "r");
+    const body = out.filter((l) => l.startsWith("  L"));
+    expect(body.length).toBe(LIVE_CONTENT_MAX_LINES);
+    expect(out.some((l) => l.includes("more line"))).toBe(true);
+    expect(out.some((l) => l.startsWith("  meta: "))).toBe(true);
+  });
+});
+
+describe("parseLiveArgv", () => {
+  test("parses thread id and flags in any order", () => {
+    const a = parseLiveArgv(["01ABC", "--debug", "--role", "planner"]);
+    expect(a.ok).toBe(true);
+    if (a.ok) {
+      expect(a.value.threadId).toBe("01ABC");
+      expect(a.value.latest).toBe(false);
+      expect(a.value.debug).toBe(true);
+      expect(a.value.role).toBe("planner");
+    }
+    const b = parseLiveArgv(["--latest", "--role", "x"]);
+    expect(b.ok).toBe(true);
+    if (b.ok) {
+      expect(b.value.latest).toBe(true);
+      expect(b.value.threadId).toBe(null);
+      expect(b.value.role).toBe("x");
+    }
+  });
+
+  test("rejects --latest with thread id", () => {
+    const r = parseLiveArgv(["--latest", "01ABC"]);
+    expect(r.ok).toBe(false);
+  });
+});
+
+describe("live CLI", () => {
+  let prevEnv: string | undefined;
+  let storageRoot: string;
+
+  beforeEach(async () => {
+    prevEnv = process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    storageRoot = await mkdtemp(join(tmpdir(), "uncaged-wf-live-"));
+    process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = storageRoot;
+    await mkdir(join(storageRoot, "logs", "C9NMV6V2TQT81"), { recursive: true });
+    await cp(
+      join(fixtureRoot, "logs", "C9NMV6V2TQT81", "01LIVECMPLT01DDDDDDDDDDDDG.data.jsonl"),
+      join(storageRoot, "logs", "C9NMV6V2TQT81", "01LIVECMPLT01DDDDDDDDDDDDG.data.jsonl"),
+    );
+    await cp(
+      join(fixtureRoot, "logs", "C9NMV6V2TQT81", "01LIVECMPLT01DDDDDDDDDDDDG.info.jsonl"),
+      join(storageRoot, "logs", "C9NMV6V2TQT81", "01LIVECMPLT01DDDDDDDDDDDDG.info.jsonl"),
+    );
+    await cp(
+      join(fixtureRoot, "logs", "C9NMV6V2TQT81", "01LIVEINFLY01DDDDDDDDDDDDG.data.jsonl"),
+      join(storageRoot, "logs", "C9NMV6V2TQT81", "01LIVEINFLY01DDDDDDDDDDDDG.data.jsonl"),
+    );
+    await cp(
+      join(fixtureRoot, "logs", "C9NMV6V2TQT81", "01LIVEOLDER01DDDDDDDDDDDDG.data.jsonl"),
+      join(storageRoot, "logs", "C9NMV6V2TQT81", "01LIVEOLDER01DDDDDDDDDDDDG.data.jsonl"),
+    );
+
+    const cas = createCasStore(getGlobalCasDir(storageRoot));
+    await putContentMerkleNode(cas, LIVE_FIXTURE_PLANNER_BODY);
+    await putContentMerkleNode(cas, "patch");
+    await putContentMerkleNode(cas, "still running");
+  });
+
+  afterEach(async () => {
+    if (prevEnv === undefined) {
+      delete process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    } else {
+      process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = prevEnv;
+    }
+    await rm(storageRoot, { recursive: true, force: true });
+  });
+
+  test("prints role steps and summary for a completed thread", async () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const proc = spawn(process.execPath, [cliEntryPath, "live", "01LIVECMPLT01DDDDDDDDDDDDG"], {
+      env,
+      stdio: ["ignore", "pipe", "pipe"],
+    });
+    const stdout = await new Promise<string>((resolve, reject) => {
+      let buf = "";
+      proc.stdout?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.stderr?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.on("error", reject);
+      proc.on("exit", (code: number | null) => {
+        if (code === 0) {
+          resolve(buf);
+        } else {
+          reject(new Error(`exit ${code}: ${buf}`));
+        }
+      });
+    });
+
+    expect(stdout).toContain("planner");
+    expect(stdout).toContain("coder");
+    expect(stdout).toContain("meta:");
+    expect(stdout).toContain('"phase":"plan"');
+    expect(stdout).toContain("LINE10");
+    expect(stdout).not.toContain("LINE11");
+    expect(stdout).toContain("more line");
+    expect(stdout).toContain("completed: returnCode=0");
+    expect(stdout).toContain("fixture completed");
+  });
+
+  test("--latest tails the newest thread by start timestamp", async () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const proc = spawn(process.execPath, [cliEntryPath, "live", "--latest"], {
+      env,
+      stdio: ["ignore", "pipe", "pipe"],
+    });
+    const stdout = await new Promise<string>((resolve, reject) => {
+      let buf = "";
+      proc.stdout?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.stderr?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.on("error", reject);
+      proc.on("exit", (code: number | null) => {
+        if (code === 0) {
+          resolve(buf);
+        } else {
+          reject(new Error(`exit ${code}: ${buf}`));
+        }
+      });
+    });
+
+    expect(stdout).toContain("fixture completed");
+    expect(stdout).not.toContain("older thread");
+  });
+
+  test("--debug prints .info.jsonl records after data output", async () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const proc = spawn(
+      process.execPath,
+      [cliEntryPath, "live", "01LIVECMPLT01DDDDDDDDDDDDG", "--debug"],
+      {
+        env,
+        stdio: ["ignore", "pipe", "pipe"],
+      },
+    );
+    const stdout = await new Promise<string>((resolve, reject) => {
+      let buf = "";
+      proc.stdout?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.stderr?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.on("error", reject);
+      proc.on("exit", (code: number | null) => {
+        if (code === 0) {
+          resolve(buf);
+        } else {
+          reject(new Error(`exit ${code}: ${buf}`));
+        }
+      });
+    });
+
+    expect(stdout).toContain("[DEBUGTAG1]");
+    expect(stdout).toContain("bundle loaded");
+    expect(stdout).toContain("[DEBUGTAG2]");
+    expect(stdout).toContain("multi line");
+  });
+
+  test("--role filters out non-matching roles", async () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const proc = spawn(
+      process.execPath,
+      [cliEntryPath, "live", "01LIVECMPLT01DDDDDDDDDDDDG", "--role", "planner"],
+      {
+        env,
+        stdio: ["ignore", "pipe", "pipe"],
+      },
+    );
+    const stdout = await new Promise<string>((resolve, reject) => {
+      let buf = "";
+      proc.stdout?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.stderr?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.on("error", reject);
+      proc.on("exit", (code: number | null) => {
+        if (code === 0) {
+          resolve(buf);
+        } else {
+          reject(new Error(`exit ${code}: ${buf}`));
+        }
+      });
+    });
+
+    expect(stdout).toContain("planner");
+    expect(stdout).not.toContain("patch");
+    expect(stdout).toContain("completed: returnCode=0");
+  });
+
+  test("--latest --debug --role combine", async () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const proc = spawn(
+      process.execPath,
+      [cliEntryPath, "live", "--latest", "--debug", "--role", "planner"],
+      {
+        env,
+        stdio: ["ignore", "pipe", "pipe"],
+      },
+    );
+    const stdout = await new Promise<string>((resolve, reject) => {
+      let buf = "";
+      proc.stdout?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.stderr?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.on("error", reject);
+      proc.on("exit", (code: number | null) => {
+        if (code === 0) {
+          resolve(buf);
+        } else {
+          reject(new Error(`exit ${code}: ${buf}`));
+        }
+      });
+    });
+
+    expect(stdout).toContain("[DEBUGTAG1]");
+    expect(stdout).toContain("planner");
+    expect(stdout).not.toContain("patch");
+    expect(stdout).toContain("fixture completed");
+  });
+
+  test("unknown thread id exits 1", () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const r = spawnSync(process.execPath, [cliEntryPath, "live", "01UNKNOWNXXXXXXXXXXXXXXXXX"], {
+      env,
+      encoding: "utf8",
+    });
+    expect(r.status).toBe(1);
+    expect(String(r.stderr ?? "")).toContain("thread not found");
+  });
+
+  test("follows file until WorkflowResult is appended", async () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
+    const dataPath = join(
+      storageRoot,
+      "logs",
+      "C9NMV6V2TQT81",
+      "01LIVEINFLY01DDDDDDDDDDDDG.data.jsonl",
+    );
+
+    const proc = spawn(process.execPath, [cliEntryPath, "live", "01LIVEINFLY01DDDDDDDDDDDDG"], {
+      env,
+      stdio: ["ignore", "pipe", "pipe"],
+    });
+
+    await new Promise((r) => setTimeout(r, 120));
+    const prior = await readFile(dataPath, "utf8");
+    await writeFile(
+      dataPath,
+      `${prior.replace(/\s*$/, "")}\n${JSON.stringify({ returnCode: 0, summary: "caught up" })}\n`,
+      "utf8",
+    );
+
+    const stdout = await new Promise<string>((resolve, reject) => {
+      let buf = "";
+      proc.stdout?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.stderr?.on("data", (c: Buffer) => {
+        buf += c.toString("utf8");
+      });
+      proc.on("error", reject);
+      proc.on("exit", (code: number | null) => {
+        if (code === 0) {
+          resolve(buf);
+        } else {
+          reject(new Error(`exit ${code}: ${buf}`));
+        }
+      });
+    });
+
+    expect(stdout).toContain("planner");
+    expect(stdout).toContain("completed: returnCode=0");
+    expect(stdout).toContain("caught up");
+  });
+});
+
+describe("live --latest with empty storage", () => {
+  let prevEnv: string | undefined;
+  let emptyRoot: string;
+
+  beforeEach(async () => {
+    prevEnv = process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    emptyRoot = await mkdtemp(join(tmpdir(), "uncaged-wf-live-empty-"));
+    process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = emptyRoot;
+  });
+
+  afterEach(async () => {
+    if (prevEnv === undefined) {
+      delete process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    } else {
+      process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = prevEnv;
+    }
+    await rm(emptyRoot, { recursive: true, force: true });
+  });
+
+  test("exits 1 when no threads exist", () => {
+    const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: emptyRoot };
+    const r = spawnSync(process.execPath, [cliEntryPath, "live", "--latest"], {
+      env,
+      encoding: "utf8",
+    });
+    expect(r.status).toBe(1);
+    expect(String(r.stderr ?? "")).toContain("no threads");
+  });
+});

packages/cli-workflow/__tests__/storage-env.test.ts

@@ -0,0 +1,54 @@
+import { afterEach, beforeEach, describe, expect, test } from "bun:test";
+import { getDefaultWorkflowStorageRoot } from "@uncaged/workflow";
+import { resolveWorkflowStorageRoot } from "../src/storage-env.js";
+
+describe("resolveWorkflowStorageRoot", () => {
+  let savedInternal: string | undefined;
+  let savedUser: string | undefined;
+
+  beforeEach(() => {
+    savedInternal = process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    savedUser = process.env.WORKFLOW_STORAGE_ROOT;
+    delete process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    delete process.env.WORKFLOW_STORAGE_ROOT;
+  });
+
+  afterEach(() => {
+    if (savedInternal === undefined) {
+      delete process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+    } else {
+      process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = savedInternal;
+    }
+    if (savedUser === undefined) {
+      delete process.env.WORKFLOW_STORAGE_ROOT;
+    } else {
+      process.env.WORKFLOW_STORAGE_ROOT = savedUser;
+    }
+  });
+
+  test("returns default when no env vars are set", () => {
+    expect(resolveWorkflowStorageRoot()).toBe(getDefaultWorkflowStorageRoot());
+  });
+
+  test("WORKFLOW_STORAGE_ROOT overrides default", () => {
+    process.env.WORKFLOW_STORAGE_ROOT = "/tmp/custom-storage";
+    expect(resolveWorkflowStorageRoot()).toBe("/tmp/custom-storage");
+  });
+
+  test("UNCAGED_WORKFLOW_STORAGE_ROOT takes priority over WORKFLOW_STORAGE_ROOT", () => {
+    process.env.WORKFLOW_STORAGE_ROOT = "/tmp/user-path";
+    process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = "/tmp/internal-path";
+    expect(resolveWorkflowStorageRoot()).toBe("/tmp/internal-path");
+  });
+
+  test("ignores empty WORKFLOW_STORAGE_ROOT", () => {
+    process.env.WORKFLOW_STORAGE_ROOT = "";
+    expect(resolveWorkflowStorageRoot()).toBe(getDefaultWorkflowStorageRoot());
+  });
+
+  test("ignores empty UNCAGED_WORKFLOW_STORAGE_ROOT and falls through to WORKFLOW_STORAGE_ROOT", () => {
+    process.env.UNCAGED_WORKFLOW_STORAGE_ROOT = "";
+    process.env.WORKFLOW_STORAGE_ROOT = "/tmp/user-fallback";
+    expect(resolveWorkflowStorageRoot()).toBe("/tmp/user-fallback");
+  });
+});

packages/cli-workflow/__tests__/thread-cli.test.ts

@@ -250,13 +250,16 @@ describe("cli thread commands", () => {
 
   test("cli entrypoint dispatches threads / ps (spawn)", () => {
     const env = { ...process.env, UNCAGED_WORKFLOW_STORAGE_ROOT: storageRoot };
-    const threads = spawnSync(process.execPath, [cliEntryPath, "threads"], {
+    const threads = spawnSync(process.execPath, [cliEntryPath, "thread", "list"], {
       env,
       encoding: "utf8",
     });
     expect(threads.status).toBe(0);
 
-    const ps = spawnSync(process.execPath, [cliEntryPath, "ps"], { env, encoding: "utf8" });
+    const ps = spawnSync(process.execPath, [cliEntryPath, "thread", "ps"], {
+      env,
+      encoding: "utf8",
+    });
     expect(ps.status).toBe(0);
   });
 
@@ -323,7 +326,7 @@ describe("cli thread commands", () => {
       .trim()
       .split("\n")
       .filter((l) => l !== "");
-    expect(lines.length).toBe(2);
+    expect(lines.length).toBe(3);
 
     const runningPath = join(dirname(dataPath), `${threadId}.running`);
     expect(await pathExists(runningPath)).toBe(false);
@@ -362,8 +365,8 @@ describe("cli thread commands", () => {
     const resumed = await cmdResume(storageRoot, threadId);
     expect(resumed.ok).toBe(true);
 
-    await waitUntilMinDataLines(dataPath, 3, 120);
-    expect(await countDataJsonlLines(dataPath)).toBe(3);
+    await waitUntilMinDataLines(dataPath, 4, 120);
+    expect(await countDataJsonlLines(dataPath)).toBe(4);
 
     const runningPath = join(dirname(dataPath), `${threadId}.running`);
     await waitUntilRunningFileAbsent(runningPath, 100);

packages/cli-workflow/src/cli-dispatch.ts

@@ -3,10 +3,12 @@ import { cmdAdd, formatAddSuccess, parseAddArgv } from "./cmd-add.js";
 import { cmdCasGet, cmdCasList, cmdCasPut, cmdCasRm } from "./cmd-cas.js";
 import { cmdFork, parseForkArgv } from "./cmd-fork.js";
 import { cmdGc } from "./cmd-gc.js";
+import { formatSkillDoc } from "./cmd-help.js";
 import { cmdHistory } from "./cmd-history.js";
 import { cmdInitTemplate, cmdInitWorkspace } from "./cmd-init.js";
 import { cmdKill } from "./cmd-kill.js";
 import { cmdList, formatListLines } from "./cmd-list.js";
+import { cmdLive } from "./cmd-live.js";
 import { cmdPause } from "./cmd-pause.js";
 import { cmdPs } from "./cmd-ps.js";
 import { cmdRemove } from "./cmd-remove.js";
@@ -16,66 +18,52 @@ import { cmdRun } from "./cmd-run.js";
 import { cmdShow, formatShowYaml } from "./cmd-show.js";
 import { cmdThreadRemove, cmdThreadShow } from "./cmd-thread.js";
 import { cmdThreads } from "./cmd-threads.js";
+import { parseLiveArgv } from "./live-argv.js";
 import { parseRunArgv } from "./run-argv.js";
 
-export function formatCliUsage(): string {
-  return [
-    "Usage:",
-    "  uncaged-workflow add <name> <file.esm.js> [--types <path>]",
-    "  uncaged-workflow list",
-    "  uncaged-workflow show <name>",
-    "  uncaged-workflow remove <name>",
-    "  uncaged-workflow run <name> [--prompt <text>] [--max-rounds N]",
-    "  uncaged-workflow ps",
-    "  uncaged-workflow kill <thread-id>",
-    "  uncaged-workflow history <name>",
-    "  uncaged-workflow rollback <name> [hash]",
-    "  uncaged-workflow pause <thread-id>",
-    "  uncaged-workflow resume <thread-id>",
-    "  uncaged-workflow threads [name]",
-    "  uncaged-workflow thread <id>",
-    "  uncaged-workflow thread rm <id>",
-    "  uncaged-workflow fork <thread-id> [--from-role <role>]",
-    "  uncaged-workflow gc",
-    "  uncaged-workflow cas get <thread-id> <hash>",
-    "  uncaged-workflow cas put <thread-id> <content>",
-    "  uncaged-workflow cas list <thread-id>",
-    "  uncaged-workflow cas rm <thread-id> <hash>",
-    "  uncaged-workflow init workspace <name>",
-    "  uncaged-workflow init template <name>",
-  ].join("\n");
-}
+type DispatchFn = (storageRoot: string, argv: string[]) => Promise<number>;
 
-async function dispatchInit(_storageRoot: string, argv: string[]): Promise<number> {
-  const sub = argv[0];
-  const name = argv[1];
-  if (sub === undefined || name === undefined || argv.length > 2) {
-    printCliError(`${formatCliUsage()}\n\nerror: init requires workspace|template <name>`);
+type CommandEntry = {
+  handler: DispatchFn;
+  args: string;
+  description: string;
+};
+
+type CommandGroup = {
+  name: string;
+  commands: ReadonlyArray<{ name: string; args: string; description: string }>;
+};
+
+// ── Individual dispatch functions ──────────────────────────────────────
+
+async function dispatchInitWorkspace(_storageRoot: string, argv: string[]): Promise<number> {
+  const name = argv[0];
+  if (name === undefined || argv.length > 1) {
+    printCliError(`${formatCliUsage()}\n\nerror: init workspace requires <name>`);
     return 1;
   }
-
-  if (sub === "workspace") {
-    const result = await cmdInitWorkspace(process.cwd(), name);
-    if (!result.ok) {
-      printCliError(result.error);
-      return 1;
-    }
-    printCliLine(`initialized workflow workspace at ${result.value.rootPath}`);
-    return 0;
+  const result = await cmdInitWorkspace(process.cwd(), name);
+  if (!result.ok) {
+    printCliError(result.error);
+    return 1;
   }
+  printCliLine(`initialized workflow workspace at ${result.value.rootPath}`);
+  return 0;
+}
 
-  if (sub === "template") {
-    const result = await cmdInitTemplate(process.cwd(), name);
-    if (!result.ok) {
-      printCliError(result.error);
-      return 1;
-    }
-    printCliLine(`initialized template at ${result.value.templatePath}`);
-    return 0;
+async function dispatchInitTemplate(_storageRoot: string, argv: string[]): Promise<number> {
+  const name = argv[0];
+  if (name === undefined || argv.length > 1) {
+    printCliError(`${formatCliUsage()}\n\nerror: init template requires <name>`);
+    return 1;
   }
-
-  printCliError(`${formatCliUsage()}\n\nerror: unknown init subcommand: ${sub}`);
-  return 1;
+  const result = await cmdInitTemplate(process.cwd(), name);
+  if (!result.ok) {
+    printCliError(result.error);
+    return 1;
+  }
+  printCliLine(`initialized template at ${result.value.templatePath}`);
+  return 0;
 }
 
 async function dispatchAdd(storageRoot: string, argv: string[]): Promise<number> {
@@ -190,6 +178,15 @@ async function dispatchKill(storageRoot: string, argv: string[]): Promise<number
   return 0;
 }
 
+async function dispatchLive(storageRoot: string, argv: string[]): Promise<number> {
+  const parsed = parseLiveArgv(argv);
+  if (!parsed.ok) {
+    printCliError(`${formatCliUsage()}\n\nerror: ${parsed.error}`);
+    return 1;
+  }
+  return cmdLive(storageRoot, parsed.value);
+}
+
 async function dispatchHistory(storageRoot: string, argv: string[]): Promise<number> {
   const name = argv[0];
   if (name === undefined || argv.length > 1) {
@@ -253,7 +250,7 @@ async function dispatchResume(storageRoot: string, argv: string[]): Promise<numb
   return 0;
 }
 
-async function dispatchThreads(storageRoot: string, argv: string[]): Promise<number> {
+async function dispatchThreadList(storageRoot: string, argv: string[]): Promise<number> {
   const result = await cmdThreads(storageRoot, argv);
   if (!result.ok) {
     printCliError(result.error);
@@ -265,10 +262,10 @@ async function dispatchThreads(storageRoot: string, argv: string[]): Promise<num
   return 0;
 }
 
-async function dispatchThread(storageRoot: string, argv: string[]): Promise<number> {
+async function dispatchThreadShow(storageRoot: string, argv: string[]): Promise<number> {
   const id = argv[0];
   if (id === undefined || argv.length > 1) {
-    printCliError(`${formatCliUsage()}\n\nerror: thread requires <id>`);
+    printCliError(`${formatCliUsage()}\n\nerror: thread show requires <id>`);
     return 1;
   }
   const result = await cmdThreadShow(storageRoot, id);
@@ -295,14 +292,6 @@ async function dispatchThreadRm(storageRoot: string, argv: string[]): Promise<nu
   return 0;
 }
 
-async function dispatchThreadBranch(storageRoot: string, rest: string[]): Promise<number> {
-  const sub = rest[0];
-  if (sub === "rm") {
-    return dispatchThreadRm(storageRoot, rest.slice(1));
-  }
-  return dispatchThread(storageRoot, rest);
-}
-
 async function dispatchGc(storageRoot: string, argv: string[]): Promise<number> {
   if (argv.length > 0) {
     printCliError(`${formatCliUsage()}\n\nerror: gc takes no arguments`);
@@ -335,6 +324,8 @@ async function dispatchFork(storageRoot: string, argv: string[]): Promise<number
   return 0;
 }
 
+// ── CAS subcommand table ───────────────────────────────────────────────
+
 async function dispatchCasGet(storageRoot: string, rest: string[]): Promise<number> {
   const threadId = rest[0];
   const hash = rest[1];
@@ -400,50 +391,279 @@ async function dispatchCasRm(storageRoot: string, rest: string[]): Promise<numbe
   return 0;
 }
 
-const CAS_SUBCOMMAND_TABLE: Record<
-  string,
-  (storageRoot: string, rest: string[]) => Promise<number>
-> = {
-  get: dispatchCasGet,
-  put: dispatchCasPut,
-  list: dispatchCasList,
-  rm: dispatchCasRm,
+// ── Subcommand tables with metadata ────────────────────────────────────
+
+const WORKFLOW_SUBCOMMAND_TABLE: Record<string, CommandEntry> = {
+  add: {
+    handler: dispatchAdd,
+    args: "<name> <file.esm.js> [--types <path>]",
+    description: "Register a workflow bundle in the registry",
+  },
+  list: { handler: dispatchList, args: "", description: "List all registered workflows" },
+  show: {
+    handler: dispatchShow,
+    args: "<name>",
+    description: "Show details of a registered workflow",
+  },
+  rm: {
+    handler: dispatchRemove,
+    args: "<name>",
+    description: "Remove a workflow from the registry",
+  },
+  history: {
+    handler: dispatchHistory,
+    args: "<name>",
+    description: "Show version history of a workflow",
+  },
+  rollback: {
+    handler: dispatchRollback,
+    args: "<name> [hash]",
+    description: "Rollback a workflow to a previous version",
+  },
 };
 
-async function dispatchCas(storageRoot: string, argv: string[]): Promise<number> {
-  const sub = argv[0];
-  if (sub === undefined) {
-    printCliError(`${formatCliUsage()}\n\nerror: unknown cas subcommand: (none)`);
-    return 1;
-  }
-  const handler = CAS_SUBCOMMAND_TABLE[sub];
-  if (handler === undefined) {
-    printCliError(`${formatCliUsage()}\n\nerror: unknown cas subcommand: ${sub}`);
-    return 1;
-  }
-  return handler(storageRoot, argv.slice(1));
+const THREAD_SUBCOMMAND_TABLE: Record<string, CommandEntry> = {
+  run: {
+    handler: dispatchRun,
+    args: "<name> [--prompt <text>] [--max-rounds N]",
+    description: "Start a new thread executing a workflow",
+  },
+  list: {
+    handler: dispatchThreadList,
+    args: "[name]",
+    description: "List threads, optionally filtered by workflow name",
+  },
+  show: { handler: dispatchThreadShow, args: "<id>", description: "Show thread details and state" },
+  rm: { handler: dispatchThreadRm, args: "<id>", description: "Remove a thread" },
+  fork: {
+    handler: dispatchFork,
+    args: "<thread-id> [--from-role <role>]",
+    description: "Fork a thread, optionally from a specific role",
+  },
+  ps: { handler: dispatchPs, args: "", description: "List running threads" },
+  kill: { handler: dispatchKill, args: "<thread-id>", description: "Kill a running thread" },
+  live: {
+    handler: dispatchLive,
+    args: "<thread-id> [--debug] [--role <name>]",
+    description: "Attach to a thread and stream output live",
+  },
+  pause: { handler: dispatchPause, args: "<thread-id>", description: "Pause a running thread" },
+  resume: { handler: dispatchResume, args: "<thread-id>", description: "Resume a paused thread" },
+};
+
+const CAS_SUBCOMMAND_TABLE: Record<string, CommandEntry> = {
+  get: {
+    handler: dispatchCasGet,
+    args: "<thread-id> <hash>",
+    description: "Retrieve content by hash from a thread's CAS",
+  },
+  put: {
+    handler: dispatchCasPut,
+    args: "<thread-id> <content>",
+    description: "Store content in a thread's CAS, returns hash",
+  },
+  list: {
+    handler: dispatchCasList,
+    args: "<thread-id>",
+    description: "List all CAS entries for a thread",
+  },
+  rm: { handler: dispatchCasRm, args: "<thread-id> <hash>", description: "Remove a CAS entry" },
+  gc: { handler: dispatchGc, args: "", description: "Garbage-collect unreferenced CAS entries" },
+};
+
+const INIT_SUBCOMMAND_TABLE: Record<string, CommandEntry> = {
+  workspace: {
+    handler: dispatchInitWorkspace,
+    args: "<name>",
+    description: "Initialize a new workflow workspace",
+  },
+  template: {
+    handler: dispatchInitTemplate,
+    args: "<name>",
+    description: "Initialize a new workflow template",
+  },
+};
+
+// ── Command registry ───────────────────────────────────────────────────
+
+export function getCommandRegistry(): ReadonlyArray<CommandGroup> {
+  return [
+    {
+      name: "workflow",
+      commands: Object.entries(WORKFLOW_SUBCOMMAND_TABLE).map(([name, e]) => ({
+        name,
+        args: e.args,
+        description: e.description,
+      })),
+    },
+    {
+      name: "thread",
+      commands: Object.entries(THREAD_SUBCOMMAND_TABLE).map(([name, e]) => ({
+        name,
+        args: e.args,
+        description: e.description,
+      })),
+    },
+    {
+      name: "cas",
+      commands: Object.entries(CAS_SUBCOMMAND_TABLE).map(([name, e]) => ({
+        name,
+        args: e.args,
+        description: e.description,
+      })),
+    },
+    {
+      name: "init",
+      commands: Object.entries(INIT_SUBCOMMAND_TABLE).map(([name, e]) => ({
+        name,
+        args: e.args,
+        description: e.description,
+      })),
+    },
+  ];
 }
 
-type DispatchFn = (storageRoot: string, argv: string[]) => Promise<number>;
+// ── Auto-generated CLI usage ───────────────────────────────────────────
+
+export function formatCliUsage(): string {
+  const groups = getCommandRegistry();
+  const lines: string[] = ["Usage:"];
+  for (const group of groups) {
+    for (const cmd of group.commands) {
+      const args = cmd.args ? ` ${cmd.args}` : "";
+      lines.push(`  uncaged-workflow ${group.name} ${cmd.name}${args}`);
+    }
+    lines.push("");
+  }
+  lines.push("  uncaged-workflow run <name> [...]          (shortcut for thread run)");
+  lines.push("  uncaged-workflow live <thread-id> [...]    (shortcut for thread live)");
+  lines.push("");
+  lines.push("Environment variables:");
+  lines.push(
+    "  WORKFLOW_STORAGE_ROOT              Override storage directory (default: ~/.uncaged/workflow)",
+  );
+  lines.push(
+    "  UNCAGED_WORKFLOW_STORAGE_ROOT      Internal override (takes priority over WORKFLOW_STORAGE_ROOT)",
+  );
+  return lines.join("\n");
+}
+
+function printDeprecation(oldCmd: string, newCmd: string): void {
+  printCliWarn(`⚠ "${oldCmd}" is deprecated, use "${newCmd}" instead`);
+}
+
+// ── Group dispatchers ──────────────────────────────────────────────────
+
+function dispatchGroup(
+  tableName: string,
+  table: Record<string, CommandEntry>,
+  storageRoot: string,
+  argv: string[],
+): Promise<number> | null {
+  const sub = argv[0];
+  if (sub === undefined) {
+    printCliError(`${formatCliUsage()}\n\nerror: unknown ${tableName} subcommand: (none)`);
+    return Promise.resolve(1);
+  }
+  const entry = table[sub];
+  if (entry === undefined) {
+    return null;
+  }
+  return entry.handler(storageRoot, argv.slice(1));
+}
+
+async function dispatchInit(storageRoot: string, argv: string[]): Promise<number> {
+  const sub = argv[0];
+  const name = argv[1];
+  if (sub === undefined || name === undefined || argv.length > 2) {
+    printCliError(`${formatCliUsage()}\n\nerror: init requires workspace|template <name>`);
+    return 1;
+  }
+
+  const entry = INIT_SUBCOMMAND_TABLE[sub];
+  if (entry !== undefined) {
+    return entry.handler(storageRoot, argv.slice(1));
+  }
+
+  printCliError(`${formatCliUsage()}\n\nerror: unknown init subcommand: ${sub}`);
+  return 1;
+}
+
+async function dispatchWorkflow(storageRoot: string, argv: string[]): Promise<number> {
+  const result = dispatchGroup("workflow", WORKFLOW_SUBCOMMAND_TABLE, storageRoot, argv);
+  if (result !== null) {
+    return result;
+  }
+  const sub = argv[0];
+  if (sub === "remove") {
+    printDeprecation("workflow remove", "workflow rm");
+    return dispatchRemove(storageRoot, argv.slice(1));
+  }
+  printCliError(`${formatCliUsage()}\n\nerror: unknown workflow subcommand: ${sub}`);
+  return 1;
+}
+
+async function dispatchThread(storageRoot: string, argv: string[]): Promise<number> {
+  const result = dispatchGroup("thread", THREAD_SUBCOMMAND_TABLE, storageRoot, argv);
+  if (result !== null) {
+    return result;
+  }
+  const sub = argv[0];
+  printCliError(`${formatCliUsage()}\n\nerror: unknown thread subcommand: ${sub}`);
+  return 1;
+}
+
+async function dispatchCas(storageRoot: string, argv: string[]): Promise<number> {
+  const result = dispatchGroup("cas", CAS_SUBCOMMAND_TABLE, storageRoot, argv);
+  if (result !== null) {
+    return result;
+  }
+  const sub = argv[0];
+  printCliError(`${formatCliUsage()}\n\nerror: unknown cas subcommand: ${sub}`);
+  return 1;
+}
+
+// ── Help ────────────────────────────────────────────────────────────────
+
+async function dispatchHelp(_storageRoot: string, argv: string[]): Promise<number> {
+  if (argv.includes("--skill")) {
+    printCliLine(formatSkillDoc());
+  } else {
+    printCliLine(formatCliUsage());
+  }
+  return 0;
+}
+
+// ── Top-level command table (Phase 3) ──────────────────────────────────
 
 const COMMAND_TABLE: Record<string, DispatchFn> = {
-  add: dispatchAdd,
-  init: dispatchInit,
-  list: dispatchList,
-  show: dispatchShow,
-  remove: dispatchRemove,
-  run: dispatchRun,
-  ps: dispatchPs,
-  kill: dispatchKill,
-  history: dispatchHistory,
-  rollback: dispatchRollback,
-  pause: dispatchPause,
-  resume: dispatchResume,
-  threads: dispatchThreads,
-  thread: dispatchThreadBranch,
-  fork: dispatchFork,
-  gc: dispatchGc,
+  // Grouped commands (primary)
+  workflow: dispatchWorkflow,
+  thread: dispatchThread,
   cas: dispatchCas,
+  init: dispatchInit,
+  help: dispatchHelp,
+
+  // Top-level shortcuts (no deprecation)
+  run: dispatchRun,
+  live: dispatchLive,
+};
+
+// Deprecated flat commands that delegate to grouped commands
+const DEPRECATED_ALIASES: Record<string, { newCmd: string; handler: DispatchFn }> = {
+  add: { newCmd: "workflow add", handler: dispatchAdd },
+  list: { newCmd: "workflow list", handler: dispatchList },
+  show: { newCmd: "workflow show", handler: dispatchShow },
+  remove: { newCmd: "workflow rm", handler: dispatchRemove },
+  ps: { newCmd: "thread ps", handler: dispatchPs },
+  kill: { newCmd: "thread kill", handler: dispatchKill },
+  pause: { newCmd: "thread pause", handler: dispatchPause },
+  resume: { newCmd: "thread resume", handler: dispatchResume },
+  threads: { newCmd: "thread list", handler: dispatchThreadList },
+  fork: { newCmd: "thread fork", handler: dispatchFork },
+  gc: { newCmd: "cas gc", handler: dispatchGc },
+  history: { newCmd: "workflow history", handler: dispatchHistory },
+  rollback: { newCmd: "workflow rollback", handler: dispatchRollback },
 };
 
 export async function runCli(storageRoot: string, argv: string[]): Promise<number> {
@@ -457,10 +677,18 @@ export async function runCli(storageRoot: string, argv: string[]): Promise<numbe
     return 1;
   }
   const rest = argv.slice(1);
+
   const dispatch = COMMAND_TABLE[command];
-  if (dispatch === undefined) {
-    printCliError(`${formatCliUsage()}\n\nerror: unknown command ${command}`);
-    return 1;
+  if (dispatch !== undefined) {
+    return dispatch(storageRoot, rest);
   }
-  return dispatch(storageRoot, rest);
+
+  const deprecated = DEPRECATED_ALIASES[command];
+  if (deprecated !== undefined) {
+    printDeprecation(command, deprecated.newCmd);
+    return deprecated.handler(storageRoot, rest);
+  }
+
+  printCliError(`${formatCliUsage()}\n\nerror: unknown command ${command}`);
+  return 1;
 }

packages/cli-workflow/src/cmd-help.ts

@@ -0,0 +1,60 @@
+import { getCommandRegistry } from "./cli-dispatch.js";
+
+export function formatSkillDoc(): string {
+  const groups = getCommandRegistry();
+
+  const commandSections: string[] = [];
+  for (const group of groups) {
+    const rows = group.commands.map((cmd) => {
+      const args = cmd.args ? `\`${cmd.args}\`` : "(none)";
+      return `| \`${group.name} ${cmd.name}\` | ${args} | ${cmd.description} |`;
+    });
+    commandSections.push(
+      `### ${group.name}\n\n| Command | Args | Description |\n|---------|------|-------------|\n${rows.join("\n")}`,
+    );
+  }
+
+  return `# uncaged-workflow CLI Reference
+
+## Core Concepts
+
+| Concept | Description |
+|---------|-------------|
+| **Workflow** | A single-file ESM bundle (\`.esm.js\`) that exports \`run\` and \`descriptor\`. Identified by name and XXH64 hash. |
+| **Bundle** | The physical \`.esm.js\` file stored in the bundles directory. Immutable once written. |
+| **Thread** | A single execution of a workflow, identified by a ULID. Persists state as JSONL files. |
+| **CAS** | Content-Addressable Storage. Per-thread key-value store keyed by content hash. |
+| **Registry** | \`workflow.yaml\` — maps workflow names to their current and historical bundle hashes. |
+
+## Commands
+
+${commandSections.join("\n\n")}
+
+### Top-level shortcuts
+
+| Command | Equivalent | Description |
+|---------|------------|-------------|
+| \`run\` | \`thread run\` | Shortcut to start a thread |
+| \`live\` | \`thread live\` | Shortcut to attach to a thread |
+
+## Typical Workflow
+
+1. \`uncaged-workflow workflow add my-wf ./my-wf.esm.js\` — register a workflow
+2. \`uncaged-workflow run my-wf --prompt "do the thing"\` — start a thread
+3. \`uncaged-workflow live --latest\` — attach and watch output
+4. \`uncaged-workflow thread show <thread-id>\` — inspect completed thread
+
+## Exit Codes
+
+| Code | Meaning |
+|------|---------|
+| 0 | Success |
+| 1 | Error |
+
+## Environment Variables
+
+| Variable | Description |
+|----------|-------------|
+| \`UNCAGED_WORKFLOW_STORAGE_ROOT\` | Override the default storage directory for all workflow data |
+`;
+}

packages/cli-workflow/src/cmd-live.ts

@@ -0,0 +1,463 @@
+import { watch } from "node:fs";
+import { readFile } from "node:fs/promises";
+import { dirname, join } from "node:path";
+
+import {
+  type CasStore,
+  createCasStore,
+  getContentMerklePayload,
+  getGlobalCasDir,
+  tryParseRoleStepRecord,
+  tryParseWorkflowResultRecord,
+  type WorkflowCompletion,
+} from "@uncaged/workflow";
+
+import { printCliError, printCliLine } from "./cli-output.js";
+import { pathExists } from "./fs-utils.js";
+import type { ParsedLiveArgv } from "./live-argv.js";
+import { findLatestThreadDataPath, resolveThreadDataPath } from "./thread-scan.js";
+
+export const LIVE_CONTENT_MAX_LINES = 10;
+
+export type LiveRoleRow = {
+  role: string;
+  content: string;
+  meta: Record<string, unknown>;
+  timestamp: number;
+};
+
+export function formatLiveTimeLabel(timestampMs: number): string {
+  const d = new Date(timestampMs);
+  const hh = String(d.getHours()).padStart(2, "0");
+  const mm = String(d.getMinutes()).padStart(2, "0");
+  const ss = String(d.getSeconds()).padStart(2, "0");
+  return `${hh}:${mm}:${ss}`;
+}
+
+function shouldUseColor(): boolean {
+  return process.stdout.isTTY === true && process.env.NO_COLOR === undefined;
+}
+
+function highlightLiveRole(name: string): string {
+  if (!shouldUseColor()) {
+    return name;
+  }
+  return `\x1b[1m\x1b[36m${name}\x1b[0m`;
+}
+
+function dimGreyLine(line: string): string {
+  if (!shouldUseColor()) {
+    return line;
+  }
+  return `\x1b[2m\x1b[90m${line}\x1b[0m`;
+}
+
+export function formatLiveDebugLine(timestampMs: number, tag: string, message: string): string {
+  const label = `[${formatLiveTimeLabel(timestampMs)}] [${tag}] ${message.replace(/\n/g, " ")}`;
+  return dimGreyLine(label);
+}
+
+export function renderLiveRoleStepLines(row: LiveRoleRow, roleDisplay: string): string[] {
+  const header = `[${formatLiveTimeLabel(row.timestamp)}] ▶ ${roleDisplay}`;
+  const lines: string[] = [header];
+  const parts = row.content.split("\n");
+  const shown = parts.slice(0, LIVE_CONTENT_MAX_LINES);
+  for (const ln of shown) {
+    lines.push(`  ${ln}`);
+  }
+  const omitted = parts.length - shown.length;
+  if (omitted > 0) {
+    lines.push(`  … (${omitted} more line${omitted === 1 ? "" : "s"})`);
+  }
+  lines.push(`  meta: ${JSON.stringify(row.meta)}`);
+  return lines;
+}
+
+function printSummary(result: WorkflowCompletion): void {
+  printCliLine(`completed: returnCode=${result.returnCode} — ${result.summary}`);
+}
+
+type LiveSessionState = {
+  sawStart: boolean;
+  completed: boolean;
+  carry: string;
+  contentOffset: number;
+};
+
+type InfoLiveState = {
+  carry: string;
+  contentOffset: number;
+};
+
+function tryParseInfoRecord(obj: Record<string, unknown>): {
+  tag: string;
+  content: string;
+  timestamp: number;
+} | null {
+  const tag = obj.tag;
+  const content = obj.content;
+  const timestamp = obj.timestamp;
+  if (
+    typeof tag !== "string" ||
+    typeof content !== "string" ||
+    typeof timestamp !== "number" ||
+    !Number.isFinite(timestamp)
+  ) {
+    return null;
+  }
+  return { tag, content, timestamp };
+}
+
+async function handleJsonlLine(
+  rawLine: string,
+  state: LiveSessionState,
+  roleFilter: string | null,
+  cas: CasStore,
+): Promise<{ parseError: string | null; workflowResult: WorkflowCompletion | null }> {
+  const trimmed = rawLine.trim();
+  if (trimmed === "") {
+    return { parseError: null, workflowResult: null };
+  }
+
+  let rec: unknown;
+  try {
+    rec = JSON.parse(trimmed) as unknown;
+  } catch {
+    return { parseError: "invalid JSON in thread data file", workflowResult: null };
+  }
+  if (rec === null || typeof rec !== "object") {
+    return { parseError: "invalid record in thread data file", workflowResult: null };
+  }
+  const obj = rec as Record<string, unknown>;
+
+  if (!state.sawStart) {
+    state.sawStart = true;
+    return { parseError: null, workflowResult: null };
+  }
+
+  const wf = tryParseWorkflowResultRecord(obj);
+  if (wf !== null) {
+    state.completed = true;
+    return { parseError: null, workflowResult: wf };
+  }
+
+  const roleRow = tryParseRoleStepRecord(obj);
+  if (roleRow === null) {
+    return {
+      parseError: "unrecognized record in thread data (expected role step or result)",
+      workflowResult: null,
+    };
+  }
+
+  if (roleFilter !== null && roleRow.role !== roleFilter) {
+    return { parseError: null, workflowResult: null };
+  }
+
+  const payload = await getContentMerklePayload(cas, roleRow.contentHash);
+  const content =
+    payload !== null ? payload : `(content not in CAS; contentHash=${roleRow.contentHash})`;
+
+  const row: LiveRoleRow = {
+    role: roleRow.role,
+    content,
+    meta: roleRow.meta,
+    timestamp: roleRow.timestamp,
+  };
+  for (const outLine of renderLiveRoleStepLines(row, highlightLiveRole(row.role))) {
+    printCliLine(outLine);
+  }
+  return { parseError: null, workflowResult: null };
+}
+
+async function pumpNewContent(
+  dataPath: string,
+  state: LiveSessionState,
+  roleFilter: string | null,
+  cas: CasStore,
+): Promise<number | null> {
+  let text: string;
+  try {
+    text = await readFile(dataPath, "utf8");
+  } catch {
+    return null;
+  }
+
+  if (text.length < state.contentOffset) {
+    state.contentOffset = 0;
+    state.carry = "";
+  }
+
+  const chunk = text.slice(state.contentOffset);
+  state.contentOffset = text.length;
+  state.carry += chunk;
+
+  const parts = state.carry.split("\n");
+  state.carry = parts.pop() ?? "";
+
+  for (const line of parts) {
+    const { parseError, workflowResult } = await handleJsonlLine(line, state, roleFilter, cas);
+    if (parseError !== null) {
+      printCliError(parseError);
+      return 1;
+    }
+    if (workflowResult !== null) {
+      printSummary(workflowResult);
+      return 0;
+    }
+  }
+
+  return null;
+}
+
+async function pumpNewInfoContent(infoPath: string, state: InfoLiveState): Promise<void> {
+  let text: string;
+  try {
+    text = await readFile(infoPath, "utf8");
+  } catch {
+    return;
+  }
+
+  if (text.length < state.contentOffset) {
+    state.contentOffset = 0;
+    state.carry = "";
+  }
+
+  const chunk = text.slice(state.contentOffset);
+  state.contentOffset = text.length;
+  state.carry += chunk;
+
+  const parts = state.carry.split("\n");
+  state.carry = parts.pop() ?? "";
+
+  for (const line of parts) {
+    const trimmed = line.trim();
+    if (trimmed === "") {
+      continue;
+    }
+    let rec: unknown;
+    try {
+      rec = JSON.parse(trimmed) as unknown;
+    } catch {
+      continue;
+    }
+    if (rec === null || typeof rec !== "object") {
+      continue;
+    }
+    const parsed = tryParseInfoRecord(rec as Record<string, unknown>);
+    if (parsed === null) {
+      continue;
+    }
+    printCliLine(formatLiveDebugLine(parsed.timestamp, parsed.tag, parsed.content));
+  }
+}
+
+type WatchPumpTask = {
+  path: string;
+  pump: () => Promise<number | null>;
+};
+
+async function runWatchPumpStep(
+  settled: () => boolean,
+  pump: () => Promise<number | null>,
+  closeAll: () => void,
+  finish: (code: number) => void,
+): Promise<void> {
+  if (settled()) {
+    return;
+  }
+  try {
+    const code = await pump();
+    if (code !== null) {
+      closeAll();
+      finish(code);
+    }
+  } catch (e) {
+    closeAll();
+    throw e instanceof Error ? e : new Error(String(e));
+  }
+}
+
+function watchLivePaths(params: { tasks: WatchPumpTask[]; signal: AbortSignal }): Promise<number> {
+  const { tasks, signal } = params;
+
+  return new Promise((resolve, reject) => {
+    let settled = false;
+    const finish = (code: number): void => {
+      if (settled) {
+        return;
+      }
+      settled = true;
+      resolve(code);
+    };
+
+    const pumpChains = new Map<string, Promise<void>>();
+    for (const t of tasks) {
+      pumpChains.set(t.path, Promise.resolve());
+    }
+
+    const watchers: ReturnType<typeof watch>[] = [];
+
+    const closeAll = (): void => {
+      for (const w of watchers) {
+        w.close();
+      }
+    };
+
+    function schedulePump(path: string, pump: () => Promise<number | null>): void {
+      const prev = pumpChains.get(path) ?? Promise.resolve();
+      const next = (async () => {
+        await prev;
+        await runWatchPumpStep(() => settled, pump, closeAll, finish);
+      })();
+      pumpChains.set(path, next);
+    }
+
+    for (const { path, pump } of tasks) {
+      const watcher = watch(path, (eventType) => {
+        if (eventType === "rename") {
+          return;
+        }
+        schedulePump(path, pump);
+      });
+      watchers.push(watcher);
+      watcher.on("error", (err: Error) => {
+        closeAll();
+        reject(err);
+      });
+    }
+
+    const onAbort = (): void => {
+      closeAll();
+      finish(0);
+    };
+    signal.addEventListener("abort", onAbort, { once: true });
+
+    for (const { path, pump } of tasks) {
+      schedulePump(path, pump);
+    }
+  });
+}
+
+type LiveThreadTarget = {
+  threadId: string;
+  dataPath: string;
+};
+
+async function resolveLiveThreadTarget(
+  storageRoot: string,
+  parsed: ParsedLiveArgv,
+): Promise<LiveThreadTarget | null> {
+  if (parsed.latest) {
+    const found = await findLatestThreadDataPath(storageRoot);
+    if (found === null) {
+      printCliError("live: no threads found");
+      return null;
+    }
+    return found;
+  }
+
+  const id = parsed.threadId;
+  if (id === null) {
+    printCliError("live: internal error: missing thread id");
+    return null;
+  }
+  const resolved = await resolveThreadDataPath(storageRoot, id);
+  if (resolved === null) {
+    printCliError(`thread not found: ${id}`);
+    return null;
+  }
+  return { threadId: id, dataPath: resolved };
+}
+
+async function buildLiveWatchTasks(params: {
+  dataPath: string;
+  infoPath: string;
+  debug: boolean;
+  dataState: LiveSessionState;
+  infoState: InfoLiveState;
+  roleFilter: string | null;
+  cas: CasStore;
+}): Promise<WatchPumpTask[]> {
+  const { dataPath, infoPath, debug, dataState, infoState, roleFilter, cas } = params;
+  const tasks: WatchPumpTask[] = [
+    {
+      path: dataPath,
+      pump: () => pumpNewContent(dataPath, dataState, roleFilter, cas),
+    },
+  ];
+
+  if (debug && (await pathExists(infoPath))) {
+    tasks.push({
+      path: infoPath,
+      pump: async () => {
+        await pumpNewInfoContent(infoPath, infoState);
+        return null;
+      },
+    });
+  }
+
+  return tasks;
+}
+
+export async function cmdLive(storageRoot: string, parsed: ParsedLiveArgv): Promise<number> {
+  const target = await resolveLiveThreadTarget(storageRoot, parsed);
+  if (target === null) {
+    return 1;
+  }
+
+  const { threadId, dataPath } = target;
+  const roleFilter = parsed.role;
+  const infoPath = join(dirname(dataPath), `${threadId}.info.jsonl`);
+  const cas = createCasStore(getGlobalCasDir(storageRoot));
+
+  const dataState: LiveSessionState = {
+    sawStart: false,
+    completed: false,
+    carry: "",
+    contentOffset: 0,
+  };
+
+  const infoState: InfoLiveState = {
+    carry: "",
+    contentOffset: 0,
+  };
+
+  const controller = new AbortController();
+  const onSigInt = (): void => {
+    controller.abort();
+  };
+  process.on("SIGINT", onSigInt);
+
+  try {
+    const firstData = await pumpNewContent(dataPath, dataState, roleFilter, cas);
+    if (firstData === 1) {
+      return 1;
+    }
+
+    if (parsed.debug && (await pathExists(infoPath))) {
+      await pumpNewInfoContent(infoPath, infoState);
+    }
+
+    if (firstData === 0 || dataState.completed) {
+      return 0;
+    }
+
+    const tasks = await buildLiveWatchTasks({
+      dataPath,
+      infoPath,
+      debug: parsed.debug,
+      dataState,
+      infoState,
+      roleFilter,
+      cas,
+    });
+
+    return await watchLivePaths({ tasks, signal: controller.signal });
+  } catch (e) {
+    const message = e instanceof Error ? e.message : String(e);
+    printCliError(`live: ${message}`);
+    return 1;
+  } finally {
+    process.off("SIGINT", onSigInt);
+  }
+}

packages/cli-workflow/src/live-argv.ts

@@ -0,0 +1,75 @@
+import { err, ok, type Result } from "@uncaged/workflow";
+
+export type ParsedLiveArgv = {
+  threadId: string | null;
+  latest: boolean;
+  debug: boolean;
+  role: string | null;
+};
+
+type LiveArgvScan = {
+  latest: boolean;
+  debug: boolean;
+  role: string | null;
+  threadId: string | null;
+};
+
+function applyLiveArgvToken(argv: string[], i: number, s: LiveArgvScan): Result<number, string> {
+  const a = argv[i];
+  if (a === "--latest") {
+    s.latest = true;
+    return ok(i + 1);
+  }
+  if (a === "--debug") {
+    s.debug = true;
+    return ok(i + 1);
+  }
+  if (a === "--role") {
+    const v = argv[i + 1];
+    if (v === undefined || v.startsWith("--")) {
+      return err("missing value for --role");
+    }
+    s.role = v;
+    return ok(i + 2);
+  }
+  if (a.startsWith("--")) {
+    return err(`unknown live flag: ${a}`);
+  }
+  if (s.threadId !== null) {
+    return err("unexpected extra argument");
+  }
+  s.threadId = a;
+  return ok(i + 1);
+}
+
+export function parseLiveArgv(argv: string[]): Result<ParsedLiveArgv, string> {
+  const s: LiveArgvScan = {
+    latest: false,
+    debug: false,
+    role: null,
+    threadId: null,
+  };
+
+  let i = 0;
+  while (i < argv.length) {
+    const step = applyLiveArgvToken(argv, i, s);
+    if (!step.ok) {
+      return step;
+    }
+    i = step.value;
+  }
+
+  if (s.latest && s.threadId !== null) {
+    return err("live --latest does not take <thread-id>");
+  }
+  if (!s.latest && s.threadId === null) {
+    return err("live requires <thread-id> or --latest");
+  }
+
+  return ok({
+    threadId: s.threadId,
+    latest: s.latest,
+    debug: s.debug,
+    role: s.role,
+  });
+}

packages/cli-workflow/src/storage-env.ts

@@ -1,10 +1,21 @@
 import { getDefaultWorkflowStorageRoot } from "@uncaged/workflow";
 
-/** Resolve storage root, honoring `UNCAGED_WORKFLOW_STORAGE_ROOT` for tests/tools. */
+/**
+ * Resolve storage root with env var override support.
+ *
+ * Priority (highest first):
+ *  1. `UNCAGED_WORKFLOW_STORAGE_ROOT` — internal/test override
+ *  2. `WORKFLOW_STORAGE_ROOT` — user-facing override
+ *  3. Default (`~/.uncaged/workflow`)
+ */
 export function resolveWorkflowStorageRoot(): string {
-  const override = process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
-  if (override !== undefined && override !== "") {
-    return override;
+  const internal = process.env.UNCAGED_WORKFLOW_STORAGE_ROOT;
+  if (internal !== undefined && internal !== "") {
+    return internal;
+  }
+  const userOverride = process.env.WORKFLOW_STORAGE_ROOT;
+  if (userOverride !== undefined && userOverride !== "") {
+    return userOverride;
   }
   return getDefaultWorkflowStorageRoot();
 }

packages/cli-workflow/src/thread-scan.ts

@@ -1,4 +1,4 @@
-import { readdir } from "node:fs/promises";
+import { readdir, stat } from "node:fs/promises";
 import { join } from "node:path";
 
 import { pathExists, readTextFileIfExists } from "./fs-utils.js";
@@ -15,6 +15,28 @@ export type HistoricalThreadRow = {
   workflowName: string | null;
 };
 
+async function readThreadStartTimestampMs(dataPath: string): Promise<number | null> {
+  const text = await readTextFileIfExists(dataPath);
+  if (text === null) {
+    return null;
+  }
+  const firstLine = text.split("\n")[0];
+  if (firstLine === undefined || firstLine.trim() === "") {
+    return null;
+  }
+  let parsed: unknown;
+  try {
+    parsed = JSON.parse(firstLine) as unknown;
+  } catch {
+    return null;
+  }
+  if (parsed === null || typeof parsed !== "object") {
+    return null;
+  }
+  const ts = (parsed as Record<string, unknown>).timestamp;
+  return typeof ts === "number" && Number.isFinite(ts) ? ts : null;
+}
+
 async function readWorkflowNameFromDataJsonl(dataPath: string): Promise<string | null> {
   const text = await readTextFileIfExists(dataPath);
   if (text === null) {
@@ -124,6 +146,50 @@ export async function listHistoricalThreads(
   return out;
 }
 
+/**
+ * Picks the thread whose `.data.jsonl` is newest by start-record `timestamp`,
+ * falling back to file `mtime` when the timestamp is missing.
+ * Tie-breaker: larger `mtime` wins when start timestamps are equal.
+ */
+export async function findLatestThreadDataPath(
+  storageRoot: string,
+): Promise<{ threadId: string; dataPath: string } | null> {
+  const threads = await listHistoricalThreads(storageRoot, null);
+  if (threads.length === 0) {
+    return null;
+  }
+
+  let best: {
+    threadId: string;
+    dataPath: string;
+    primary: number;
+    secondary: number;
+  } | null = null;
+
+  for (const t of threads) {
+    const dataPath = join(storageRoot, "logs", t.hash, `${t.threadId}.data.jsonl`);
+    let mtimeMs = 0;
+    try {
+      const st = await stat(dataPath);
+      mtimeMs = st.mtimeMs;
+    } catch {
+      continue;
+    }
+    const startTs = await readThreadStartTimestampMs(dataPath);
+    const primary = startTs !== null ? startTs : mtimeMs;
+    const secondary = mtimeMs;
+    if (
+      best === null ||
+      primary > best.primary ||
+      (primary === best.primary && secondary > best.secondary)
+    ) {
+      best = { threadId: t.threadId, dataPath, primary, secondary };
+    }
+  }
+
+  return best === null ? null : { threadId: best.threadId, dataPath: best.dataPath };
+}
+
 export async function resolveThreadDataPath(
   storageRoot: string,
   threadId: string,

packages/workflow-role-committer/src/committer.ts

@@ -21,12 +21,12 @@ export const committerMetaSchema = z.discriminatedUnion("status", [
 
 export type CommitterMeta = z.infer<typeof committerMetaSchema>;
 
-const COMMITTER_SYSTEM = `You are the git committer. Create a branch, commit the changes, and push.
+const COMMITTER_SYSTEM = `You are the git committer. Create a branch and commit the changes.
 Report the branch name and commit SHA. On failure, classify as recoverable or unrecoverable.
 Do not attempt to fix failures yourself.`;
 
 export const committerRole: RoleDefinition<CommitterMeta> = {
-  description: "Creates branch, commits, and pushes when review passes.",
+  description: "Creates a branch and commits changes.",
   systemPrompt: COMMITTER_SYSTEM,
   extractPrompt:
     "Extract the commit result: committed (with branch and SHA), recoverable failure, or unrecoverable failure. Include error details and log references if applicable.",

packages/workflow-role-planner/src/planner.ts

@@ -36,6 +36,16 @@ Replace \`<THREAD_ID>\` with the actual thread ID you found above. The command p
 
 **Do NOT store phase details in any other way** (no temp files, no invented paths). The CLI command is the only supported storage mechanism.
 
+## Phase granularity
+
+Match the number of phases to task complexity:
+- Trivial (add a config option, fix a typo, rename): 1 phase
+- Small (a new feature touching 2-3 files): 1-2 phases
+- Medium (cross-module refactor): 2-3 phases
+- Large (new subsystem, architectural change): 3-5 phases
+
+Fewer phases is always better. Each phase must justify its existence — if two phases would be tested together anyway, merge them.
+
 ## Output format
 
 After storing all phases via the CLI, output compact JSON only:

packages/workflow-role-submitter/__tests__/submitter.test.ts

@@ -0,0 +1,39 @@
+import { describe, expect, test } from "bun:test";
+
+import { submitterMetaSchema, submitterRole } from "../src/submitter.js";
+
+describe("submitterRole", () => {
+  test("submitted sample validates against schema", () => {
+    const parsed = submitterMetaSchema.safeParse({
+      status: "submitted" as const,
+      prUrl: "https://github.com/example/repo/pull/42",
+    });
+    expect(parsed.success).toBe(true);
+  });
+
+  test("failed sample validates against schema", () => {
+    const parsed = submitterMetaSchema.safeParse({
+      status: "failed" as const,
+      error: "gh not authenticated",
+    });
+    expect(parsed.success).toBe(true);
+  });
+
+  test("rejects unknown status discriminant", () => {
+    const parsed = submitterMetaSchema.safeParse({
+      status: "queued",
+      prUrl: "https://example.com",
+    });
+    expect(parsed.success).toBe(false);
+  });
+
+  test("exposes submitter system prompt", () => {
+    expect(submitterRole.systemPrompt).toContain("submitter");
+    expect(submitterRole.systemPrompt).toContain("pull request");
+  });
+
+  test("uses single extract mode without refs", () => {
+    expect(submitterRole.extractMode).toBe("single");
+    expect(submitterRole.extractRefs).toBeNull();
+  });
+});

packages/workflow-role-submitter/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "@uncaged/workflow-role-submitter",
+  "version": "0.1.0",
+  "type": "module",
+  "main": "src/index.ts",
+  "types": "src/index.ts",
+  "scripts": {
+    "build": "echo 'TODO'",
+    "test": "bun test"
+  },
+  "dependencies": {
+    "@uncaged/workflow": "workspace:*",
+    "zod": "^4.0.0"
+  }
+}

packages/workflow-role-submitter/src/index.ts

@@ -0,0 +1 @@
+export { type SubmitterMeta, submitterMetaSchema, submitterRole } from "./submitter.js";

packages/workflow-role-submitter/src/submitter.ts

@@ -0,0 +1,44 @@
+import type { RoleDefinition } from "@uncaged/workflow";
+import * as z from "zod/v4";
+
+export const submitterMetaSchema = z.discriminatedUnion("status", [
+  z.object({
+    status: z.literal("submitted"),
+    prUrl: z.string(),
+  }),
+  z.object({
+    status: z.literal("failed"),
+    error: z.string(),
+  }),
+]);
+
+export type SubmitterMeta = z.infer<typeof submitterMetaSchema>;
+
+const SUBMITTER_SYSTEM = `You are the **submitter**. Your job is to push the work branch to the remote and open a pull request.
+
+## Inputs
+
+Read the thread for context:
+- The **preparer**'s output gives you the absolute repo path and the default branch (and remote URL by inspecting the repo).
+- The **developer**'s output gives you the branch name that was committed and a list of files changed plus a summary of the work.
+
+## Procedure
+
+1. \`cd\` into the repo path from the preparer's output.
+2. Push the developer's branch to the remote: \`git push -u origin <branch>\`.
+3. Open a pull request (e.g. via \`gh pr create\`) targeting the default branch. The PR title should be short and describe the change. The PR description should summarize what changed (drawing from the developer's summary and filesChanged) and reference the original issue/task if applicable.
+4. Report the resulting PR URL.
+
+On any failure (push rejected, gh not authenticated, PR creation failed, etc.), report status="failed" with a short error message. Do not retry — surface the error so the moderator can decide.`;
+
+const SUBMITTER_EXTRACT_PROMPT =
+  "Extract the submission result. status='submitted' with prUrl on success, or status='failed' with a short error message on failure.";
+
+export const submitterRole: RoleDefinition<SubmitterMeta> = {
+  description: "Pushes the developer's branch to the remote and opens a pull request.",
+  systemPrompt: SUBMITTER_SYSTEM,
+  extractPrompt: SUBMITTER_EXTRACT_PROMPT,
+  schema: submitterMetaSchema,
+  extractRefs: null,
+  extractMode: "single",
+};

packages/workflow-role-submitter/tsconfig.json

@@ -0,0 +1,10 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "rootDir": "src",
+    "outDir": "dist",
+    "composite": true
+  },
+  "include": ["src/**/*.ts"],
+  "references": [{ "path": "../workflow" }]
+}

packages/workflow-role-tester/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "@uncaged/workflow-role-tester",
+  "version": "0.1.0",
+  "type": "module",
+  "main": "src/index.ts",
+  "types": "src/index.ts",
+  "scripts": {
+    "build": "echo 'TODO'",
+    "test": "bun test"
+  },
+  "dependencies": {
+    "@uncaged/workflow": "workspace:*",
+    "zod": "^4.0.0"
+  }
+}

packages/workflow-role-tester/src/index.ts

@@ -0,0 +1 @@
+export { type TesterMeta, testerMetaSchema, testerRole } from "./tester.js";

packages/workflow-role-tester/src/tester.ts

@@ -0,0 +1,27 @@
+import type { RoleDefinition } from "@uncaged/workflow";
+import * as z from "zod/v4";
+
+export const testerMetaSchema = z.discriminatedUnion("status", [
+  z.object({
+    status: z.literal("passed"),
+    details: z.string(),
+  }),
+  z.object({
+    status: z.literal("failed"),
+    details: z.string(),
+  }),
+]);
+
+export type TesterMeta = z.infer<typeof testerMetaSchema>;
+
+const TESTER_SYSTEM = `You are a tester. Run the project's test suite, build, and lint commands. Check what commands are available from the preparer's output in the thread. Report pass/fail with details of what failed.`;
+
+export const testerRole: RoleDefinition<TesterMeta> = {
+  description: "Runs test, build, and lint commands and reports pass or fail with details.",
+  systemPrompt: TESTER_SYSTEM,
+  extractPrompt:
+    "Extract the verification result: passed with summary details, or failed with details of what broke.",
+  schema: testerMetaSchema,
+  extractRefs: null,
+  extractMode: "single",
+};

packages/workflow-role-tester/tsconfig.json

@@ -0,0 +1,10 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "rootDir": "src",
+    "outDir": "dist",
+    "composite": true
+  },
+  "include": ["src/**/*.ts"],
+  "references": [{ "path": "../workflow" }]
+}

packages/workflow-template-develop/__tests__/develop-template.test.ts

@@ -0,0 +1,260 @@
+import { describe, expect, test } from "bun:test";
+import {
+  END,
+  type ModeratorContext,
+  type RoleStep,
+  START,
+  validateWorkflowDescriptor,
+} from "@uncaged/workflow";
+
+import type { CommitterMeta } from "@uncaged/workflow-role-committer";
+import type { PlannerMeta } from "@uncaged/workflow-role-planner";
+
+import { buildDevelopDescriptor } from "../src/descriptor.js";
+import { developModerator } from "../src/index.js";
+import type { DevelopMeta } from "../src/roles.js";
+
+const DEFAULT_PHASES: PlannerMeta["phases"] = [
+  {
+    hash: "4KNMR2PX",
+    title: "Do the work",
+  },
+];
+
+function makeStart(maxRounds: number): ModeratorContext<DevelopMeta>["start"] {
+  return {
+    role: START,
+    content: "Implement the feature",
+    meta: { maxRounds },
+    timestamp: 0,
+  };
+}
+
+function makeCtx(
+  maxRounds: number,
+  steps: ModeratorContext<DevelopMeta>["steps"],
+): ModeratorContext<DevelopMeta> {
+  return {
+    threadId: "01TEST000000000000000000TR",
+    depth: 0,
+    start: makeStart(maxRounds),
+    steps,
+  };
+}
+
+function plannerStep(phases: PlannerMeta["phases"] = DEFAULT_PHASES): RoleStep<DevelopMeta> {
+  return {
+    role: "planner",
+    contentHash: "STUBHASHPLANNER001",
+    meta: { phases },
+    refs: phases.map((p) => p.hash),
+    timestamp: 1,
+  };
+}
+
+function coderStep(completedPhase = "4KNMR2PX"): RoleStep<DevelopMeta> {
+  return {
+    role: "coder",
+    contentHash: "STUBHASHCODER00001",
+    meta: { completedPhase, filesChanged: ["a.ts"], summary: "implemented" },
+    refs: [completedPhase],
+    timestamp: 2,
+  };
+}
+
+function reviewerStep(approved: boolean): RoleStep<DevelopMeta> {
+  return {
+    role: "reviewer",
+    contentHash: "STUBHASHREVIEWER01",
+    meta: approved
+      ? { status: "approved" as const }
+      : { status: "rejected" as const, issues: ["needs fix"] },
+    refs: [],
+    timestamp: 3,
+  };
+}
+
+function testerStep(passed: boolean): RoleStep<DevelopMeta> {
+  return {
+    role: "tester",
+    contentHash: "STUBHASHTESTER01",
+    meta: passed
+      ? { status: "passed" as const, details: "all checks passed" }
+      : { status: "failed" as const, details: "lint failed" },
+    refs: [],
+    timestamp: 4,
+  };
+}
+
+function committerStep(meta: CommitterMeta): RoleStep<DevelopMeta> {
+  return {
+    role: "committer",
+    contentHash: "STUBHASHCOMMITTER1",
+    meta,
+    refs: [],
+    timestamp: 5,
+  };
+}
+
+describe("developModerator", () => {
+  test("routes initial → planner → coder → reviewer → tester → committer → END", () => {
+    expect(developModerator(makeCtx(20, []))).toBe("planner");
+    expect(developModerator(makeCtx(20, [plannerStep()]))).toBe("coder");
+    expect(developModerator(makeCtx(20, [plannerStep(), coderStep()]))).toBe("reviewer");
+    expect(developModerator(makeCtx(20, [plannerStep(), coderStep(), reviewerStep(true)]))).toBe(
+      "tester",
+    );
+    expect(
+      developModerator(
+        makeCtx(20, [plannerStep(), coderStep(), reviewerStep(true), testerStep(true)]),
+      ),
+    ).toBe("committer");
+    expect(
+      developModerator(
+        makeCtx(20, [
+          plannerStep(),
+          coderStep(),
+          reviewerStep(true),
+          testerStep(true),
+          committerStep({ status: "committed", branch: "feat/x", commitSha: "abc1234" }),
+        ]),
+      ),
+    ).toBe(END);
+  });
+
+  test("reviewer rejects → coder retry when budget allows", () => {
+    const steps: ModeratorContext<DevelopMeta>["steps"] = [
+      plannerStep(),
+      coderStep(),
+      reviewerStep(false),
+    ];
+    expect(developModerator(makeCtx(20, steps))).toBe("coder");
+  });
+
+  test("reviewer rejects → END when max rounds exhausted", () => {
+    const steps: ModeratorContext<DevelopMeta>["steps"] = [
+      plannerStep(),
+      coderStep(),
+      reviewerStep(false),
+    ];
+    expect(developModerator(makeCtx(4, steps))).toBe(END);
+  });
+
+  test("tester failed → coder retry when budget allows", () => {
+    const steps: ModeratorContext<DevelopMeta>["steps"] = [
+      plannerStep(),
+      coderStep(),
+      reviewerStep(true),
+      testerStep(false),
+    ];
+    expect(developModerator(makeCtx(20, steps))).toBe("coder");
+  });
+
+  test("tester failed → END when max rounds exhausted", () => {
+    const steps: ModeratorContext<DevelopMeta>["steps"] = [
+      plannerStep(),
+      coderStep(),
+      reviewerStep(true),
+      testerStep(false),
+    ];
+    expect(developModerator(makeCtx(5, steps))).toBe(END);
+  });
+
+  test("multiple planner phases → coder until all complete, then reviewer", () => {
+    const phases: PlannerMeta["phases"] = [
+      { hash: "AA000001", title: "first phase" },
+      { hash: "AA000002", title: "second phase" },
+    ];
+    expect(developModerator(makeCtx(20, [plannerStep(phases)]))).toBe("coder");
+    expect(developModerator(makeCtx(20, [plannerStep(phases), coderStep("AA000001")]))).toBe(
+      "coder",
+    );
+    expect(
+      developModerator(
+        makeCtx(20, [plannerStep(phases), coderStep("AA000001"), coderStep("AA000002")]),
+      ),
+    ).toBe("reviewer");
+  });
+
+  test("one-shot coder reports only last phase hash → reviewer (moderator treats as all phases done)", () => {
+    const phases: PlannerMeta["phases"] = [
+      { hash: "BB000001", title: "setup branch" },
+      { hash: "BB000002", title: "write tests" },
+      { hash: "BB000003", title: "verify" },
+      { hash: "BB000004", title: "polish" },
+    ];
+    expect(developModerator(makeCtx(20, [plannerStep(phases), coderStep("BB000004")]))).toBe(
+      "reviewer",
+    );
+  });
+
+  test("unrecognised completedPhase hash → coder retry when budget allows", () => {
+    const phases: PlannerMeta["phases"] = [
+      { hash: "CC000001", title: "first phase" },
+      { hash: "CC000002", title: "second phase" },
+    ];
+    expect(developModerator(makeCtx(20, [plannerStep(phases), coderStep("all-done")]))).toBe(
+      "coder",
+    );
+  });
+
+  test("incomplete phases → END when max rounds exhausted", () => {
+    const phases: PlannerMeta["phases"] = [
+      { hash: "DD000001", title: "first phase" },
+      { hash: "DD000002", title: "second phase" },
+    ];
+    const steps: ModeratorContext<DevelopMeta>["steps"] = [
+      plannerStep(phases),
+      coderStep("DD000001"),
+    ];
+    expect(developModerator(makeCtx(3, steps))).toBe(END);
+  });
+
+  test("committer → END for any committer meta status", () => {
+    const committed = committerStep({ status: "committed", branch: "f", commitSha: "x" });
+    const recoverable = committerStep({
+      status: "recoverable",
+      error: "merge conflict",
+      logRef: null,
+    });
+    const unrecoverable = committerStep({
+      status: "unrecoverable",
+      error: "repo missing",
+      logRef: "log1",
+    });
+    const base: ModeratorContext<DevelopMeta>["steps"] = [
+      plannerStep(),
+      coderStep(),
+      reviewerStep(true),
+      testerStep(true),
+    ];
+    expect(developModerator(makeCtx(20, [...base, committed]))).toBe(END);
+    expect(developModerator(makeCtx(20, [...base, recoverable]))).toBe(END);
+    expect(developModerator(makeCtx(20, [...base, unrecoverable]))).toBe(END);
+  });
+});
+
+describe("buildDevelopDescriptor", () => {
+  test("lists all roles with schemas that validate", () => {
+    const descriptor = buildDevelopDescriptor();
+    const validated = validateWorkflowDescriptor(descriptor);
+    expect(validated.ok).toBe(true);
+    if (!validated.ok) {
+      throw new Error(validated.error);
+    }
+    expect(Object.keys(validated.value.roles).sort()).toEqual([
+      "coder",
+      "committer",
+      "planner",
+      "reviewer",
+      "tester",
+    ]);
+    for (const key of ["planner", "coder", "reviewer", "tester", "committer"] as const) {
+      const role = validated.value.roles[key];
+      expect(role).toBeDefined();
+      expect(typeof role.schema).toBe("object");
+      expect(role.schema).not.toBeNull();
+      expect(Array.isArray(role.schema)).toBe(false);
+    }
+  });
+});

packages/workflow-template-develop/package.json

@@ -0,0 +1,19 @@
+{
+  "name": "@uncaged/workflow-template-develop",
+  "version": "0.1.0",
+  "type": "module",
+  "main": "src/index.ts",
+  "types": "src/index.ts",
+  "scripts": {
+    "build": "echo 'TODO'",
+    "test": "bun test"
+  },
+  "dependencies": {
+    "@uncaged/workflow": "workspace:*",
+    "@uncaged/workflow-role-coder": "workspace:*",
+    "@uncaged/workflow-role-committer": "workspace:*",
+    "@uncaged/workflow-role-planner": "workspace:*",
+    "@uncaged/workflow-role-reviewer": "workspace:*",
+    "@uncaged/workflow-role-tester": "workspace:*"
+  }
+}

packages/workflow-template-develop/src/descriptor.ts

@@ -0,0 +1,12 @@
+import { buildDescriptor } from "@uncaged/workflow";
+
+import { developModerator } from "./moderator.js";
+import { DEVELOP_WORKFLOW_DESCRIPTION, developRoles } from "./roles.js";
+
+export function buildDevelopDescriptor() {
+  return buildDescriptor({
+    description: DEVELOP_WORKFLOW_DESCRIPTION,
+    roles: developRoles,
+    moderator: developModerator,
+  });
+}

packages/workflow-template-develop/src/index.ts

@@ -0,0 +1,60 @@
+import {
+  type AgentBinding,
+  createWorkflow,
+  type ExtractFn,
+  type LlmProvider,
+  type WorkflowDefinition,
+  type WorkflowFn,
+} from "@uncaged/workflow";
+
+import { developModerator } from "./moderator.js";
+import { DEVELOP_WORKFLOW_DESCRIPTION, type DevelopMeta, developRoles } from "./roles.js";
+
+export {
+  type CoderMeta,
+  coderMetaSchema,
+  coderRole,
+} from "@uncaged/workflow-role-coder";
+export {
+  type CommitterMeta,
+  committerMetaSchema,
+  committerRole,
+} from "@uncaged/workflow-role-committer";
+export {
+  type PlannerMeta,
+  phaseSchema,
+  plannerMetaSchema,
+  plannerRole,
+} from "@uncaged/workflow-role-planner";
+export {
+  type ReviewerMeta,
+  reviewerMetaSchema,
+  reviewerRole,
+} from "@uncaged/workflow-role-reviewer";
+export {
+  type TesterMeta,
+  testerMetaSchema,
+  testerRole,
+} from "@uncaged/workflow-role-tester";
+export { buildDevelopDescriptor } from "./descriptor.js";
+export { developModerator } from "./moderator.js";
+export {
+  DEVELOP_WORKFLOW_DESCRIPTION,
+  type DevelopMeta,
+  type DevelopRoles,
+  developRoles,
+} from "./roles.js";
+
+export const developWorkflowDefinition: WorkflowDefinition<DevelopMeta> = {
+  description: DEVELOP_WORKFLOW_DESCRIPTION,
+  roles: developRoles,
+  moderator: developModerator,
+};
+
+export function createDevelopRun(
+  binding: AgentBinding,
+  extract: ExtractFn,
+  llmProvider: LlmProvider | null,
+): WorkflowFn {
+  return createWorkflow(developWorkflowDefinition, binding, extract, llmProvider);
+}

packages/workflow-template-develop/src/moderator.ts

@@ -0,0 +1,89 @@
+import type { Moderator, ModeratorContext } from "@uncaged/workflow";
+import { END } from "@uncaged/workflow";
+
+import type { DevelopMeta } from "./roles.js";
+
+function coderFinishedAllPlannedPhases(
+  phases: ReadonlyArray<{ hash: string }>,
+  coderCompletedPhases: ReadonlyArray<string>,
+): boolean {
+  if (phases.length === 0) {
+    return true;
+  }
+  const plannedHashes = new Set(phases.map((p) => p.hash));
+  const lastHash = phases[phases.length - 1].hash;
+  const explicit = new Set(coderCompletedPhases.filter((h) => plannedHashes.has(h)));
+  if (phases.every((p) => explicit.has(p.hash))) {
+    return true;
+  }
+  if (coderCompletedPhases.some((h) => h === lastHash)) {
+    return true;
+  }
+  return false;
+}
+
+function nextAfterCoder(
+  ctx: ModeratorContext<DevelopMeta>,
+  maxRounds: number,
+): (keyof DevelopMeta & string) | typeof END {
+  const plannerStep = ctx.steps.find((s) => s.role === "planner");
+  if (plannerStep === undefined) {
+    return "reviewer";
+  }
+  const phases = plannerStep.meta.phases;
+  const coderCompletedPhases = ctx.steps
+    .filter((s) => s.role === "coder")
+    .map((s) => s.meta.completedPhase);
+  const allDone = coderFinishedAllPlannedPhases(phases, coderCompletedPhases);
+  if (allDone) {
+    return "reviewer";
+  }
+  if (ctx.steps.length < maxRounds - 1) {
+    return "coder";
+  }
+  return END;
+}
+
+export const developModerator: Moderator<DevelopMeta> = (ctx) => {
+  const maxRounds = ctx.start.meta.maxRounds;
+
+  if (ctx.steps.length === 0) {
+    return "planner";
+  }
+
+  const last = ctx.steps[ctx.steps.length - 1];
+
+  if (last.role === "planner") {
+    return "coder";
+  }
+
+  if (last.role === "coder") {
+    return nextAfterCoder(ctx, maxRounds);
+  }
+
+  if (last.role === "reviewer") {
+    if (last.meta.status === "approved") {
+      return "tester";
+    }
+    if (ctx.steps.length < maxRounds - 1) {
+      return "coder";
+    }
+    return END;
+  }
+
+  if (last.role === "tester") {
+    if (last.meta.status === "passed") {
+      return "committer";
+    }
+    if (ctx.steps.length < maxRounds - 1) {
+      return "coder";
+    }
+    return END;
+  }
+
+  if (last.role === "committer") {
+    return END;
+  }
+
+  return END;
+};

packages/workflow-template-develop/src/roles.ts

@@ -0,0 +1,29 @@
+import type { RoleDefinition } from "@uncaged/workflow";
+import { type CoderMeta, coderRole } from "@uncaged/workflow-role-coder";
+import { type CommitterMeta, committerRole } from "@uncaged/workflow-role-committer";
+import { type PlannerMeta, plannerRole } from "@uncaged/workflow-role-planner";
+import { type ReviewerMeta, reviewerRole } from "@uncaged/workflow-role-reviewer";
+import { type TesterMeta, testerRole } from "@uncaged/workflow-role-tester";
+
+export const DEVELOP_WORKFLOW_DESCRIPTION =
+  "Plan phases, implement incrementally, review, verify with tests/build/lint, and commit (planner → coder [repeat per phase] → reviewer → tester → committer).";
+
+export type DevelopMeta = {
+  planner: PlannerMeta;
+  coder: CoderMeta;
+  reviewer: ReviewerMeta;
+  tester: TesterMeta;
+  committer: CommitterMeta;
+};
+
+export type DevelopRoles = {
+  [K in keyof DevelopMeta]: RoleDefinition<DevelopMeta[K]>;
+};
+
+export const developRoles: DevelopRoles = {
+  planner: plannerRole,
+  coder: coderRole,
+  reviewer: reviewerRole,
+  tester: testerRole,
+  committer: committerRole,
+};

packages/workflow-template-develop/tsconfig.json

@@ -0,0 +1,17 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "rootDir": "src",
+    "outDir": "dist",
+    "composite": true
+  },
+  "include": ["src/**/*.ts"],
+  "references": [
+    { "path": "../workflow" },
+    { "path": "../workflow-role-coder" },
+    { "path": "../workflow-role-committer" },
+    { "path": "../workflow-role-planner" },
+    { "path": "../workflow-role-reviewer" },
+    { "path": "../workflow-role-tester" }
+  ]
+}

packages/workflow-template-solve-issue/__tests__/solve-issue-template.test.ts

@@ -12,41 +12,71 @@ import {
   validateWorkflowDescriptor,
 } from "@uncaged/workflow";
 
-import type { CoderMeta } from "@uncaged/workflow-role-coder";
-import type { PlannerMeta } from "@uncaged/workflow-role-planner";
 import type { PreparerMeta } from "@uncaged/workflow-role-preparer";
+import type { SubmitterMeta } from "@uncaged/workflow-role-submitter";
 
 import { buildSolveIssueDescriptor } from "../src/descriptor.js";
+import type { DeveloperMeta } from "../src/developer.js";
 import { createSolveIssueRun, solveIssueModerator } from "../src/index.js";
 import type { SolveIssueMeta } from "../src/roles.js";
 
-const DEFAULT_PHASES: PlannerMeta["phases"] = [
-  {
-    hash: "4KNMR2PX",
-    title: "Do the work",
-  },
-];
+function jsonResponse(payload: Record<string, unknown>): Response {
+  return new Response(JSON.stringify(payload), {
+    status: 200,
+    headers: { "Content-Type": "application/json" },
+  });
+}
 
-const EXPECT_PLANNER_META: PlannerMeta = {
-  phases: [
-    {
-      hash: "7BQST3VW",
-      title: "placeholder phase",
-    },
-  ],
-};
+function readToolListFromBody(init: RequestInit | undefined): readonly Record<string, unknown>[] {
+  if (init === undefined || init.body === undefined || init.body === null) {
+    return [];
+  }
+  const body = JSON.parse(String(init.body)) as Record<string, unknown>;
+  const tools = body.tools;
+  if (!Array.isArray(tools)) {
+    return [];
+  }
+  return tools.filter((t): t is Record<string, unknown> => t !== null && typeof t === "object");
+}
 
-const EXPECT_CODER_META: CoderMeta = {
-  completedPhase: "7BQST3VW",
-  filesChanged: [],
-  summary: "",
-};
+function singleToolName(tools: readonly Record<string, unknown>[]): string {
+  if (tools.length === 0) {
+    return "extract";
+  }
+  const fn = tools[0].function as Record<string, unknown> | undefined;
+  return typeof fn?.name === "string" ? fn.name : "extract";
+}
+
+function buildSingleModeResponse(args: Record<string, unknown>, toolName: string): Response {
+  return jsonResponse({
+    choices: [
+      {
+        message: {
+          tool_calls: [
+            {
+              type: "function",
+              function: { name: toolName, arguments: JSON.stringify(args) },
+            },
+          ],
+        },
+      },
+    ],
+  });
+}
+
+function buildReactModeResponse(args: Record<string, unknown>): Response {
+  // reactExtract accepts a plain-JSON assistant message and validates it
+  // directly against the schema, so we skip the cas_get / extract tool dance.
+  return jsonResponse({
+    choices: [{ message: { content: JSON.stringify(args) } }],
+  });
+}
 
 function installMockChatCompletions(sequence: ReadonlyArray<Record<string, unknown>>): () => void {
   const origFetch = globalThis.fetch;
   let i = 0;
   const mockFetch = async (
-    input: Parameters<typeof fetch>[0],
+    _input: Parameters<typeof fetch>[0],
     init?: RequestInit,
   ): Promise<Response> => {
     const args = sequence[i] ?? sequence[sequence.length - 1];
@@ -54,36 +84,11 @@ function installMockChatCompletions(sequence: ReadonlyArray<Record<string, unkno
       throw new Error("installMockChatCompletions: empty sequence");
     }
     i += 1;
-    void input;
-    const body = init?.body ? (JSON.parse(String(init.body)) as Record<string, unknown>) : {};
-    const tools = body.tools;
-    const firstTool =
-      Array.isArray(tools) && tools.length > 0 && tools[0] !== null && typeof tools[0] === "object"
-        ? (tools[0] as Record<string, unknown>)
-        : null;
-    const fn =
-      firstTool !== null ? (firstTool.function as Record<string, unknown> | undefined) : undefined;
-    const toolName = typeof fn?.name === "string" ? fn.name : "extract";
-    return new Response(
-      JSON.stringify({
-        choices: [
-          {
-            message: {
-              tool_calls: [
-                {
-                  type: "function",
-                  function: {
-                    name: toolName,
-                    arguments: JSON.stringify(args),
-                  },
-                },
-              ],
-            },
-          },
-        ],
-      }),
-      { status: 200, headers: { "Content-Type": "application/json" } },
-    );
+    const tools = readToolListFromBody(init);
+    if (tools.length > 1) {
+      return buildReactModeResponse(args);
+    }
+    return buildSingleModeResponse(args, singleToolName(tools));
   };
   globalThis.fetch = Object.assign(mockFetch, {
     preconnect: origFetch.preconnect.bind(origFetch),
@@ -134,152 +139,86 @@ function preparerStep(): RoleStep<SolveIssueMeta> {
   };
 }
 
-function plannerStep(phases: PlannerMeta["phases"] = DEFAULT_PHASES): RoleStep<SolveIssueMeta> {
+function developerStep(): RoleStep<SolveIssueMeta> {
   return {
-    role: "planner",
-    contentHash: "STUBHASHPLANNER001",
-    meta: { phases },
-    refs: phases.map((p) => p.hash),
+    role: "developer",
+    contentHash: "STUBHASHDEVELOPER1",
+    meta: {
+      branch: "feat/issue-1",
+      commitSha: "abc1234",
+      filesChanged: ["src/login.ts"],
+      summary: "Fixed flaky login test by stabilising async setup.",
+    },
+    refs: [],
     timestamp: 1,
   };
 }
 
-function coderStep(completedPhase = "4KNMR2PX"): RoleStep<SolveIssueMeta> {
+function submitterStep(meta: SubmitterMeta): RoleStep<SolveIssueMeta> {
   return {
-    role: "coder",
-    contentHash: "STUBHASHCODER00001",
-    meta: { completedPhase, filesChanged: ["a.ts"], summary: "fixed" },
-    refs: [completedPhase],
+    role: "submitter",
+    contentHash: "STUBHASHSUBMITTER1",
+    meta,
+    refs: [],
     timestamp: 2,
   };
 }
 
-function reviewerStep(approved: boolean): RoleStep<SolveIssueMeta> {
-  return {
-    role: "reviewer",
-    contentHash: "STUBHASHREVIEWER01",
-    meta: approved
-      ? { status: "approved" as const }
-      : { status: "rejected" as const, issues: ["needs fix"] },
-    refs: [],
-    timestamp: 3,
-  };
-}
-
-function committerStep(): RoleStep<SolveIssueMeta> {
-  return {
-    role: "committer",
-    contentHash: "STUBHASHCOMMITTER1",
-    meta: { status: "committed", branch: "feat/issue-1", commitSha: "abc1234" },
-    refs: [],
-    timestamp: 4,
-  };
-}
-
 const stubExtract = createExtract({
   baseUrl: "http://127.0.0.1:9",
   apiKey: "",
   model: "test",
 });
 
+const stubLlmProvider = {
+  baseUrl: "http://127.0.0.1:9",
+  apiKey: "",
+  model: "test",
+};
+
 describe("solveIssueModerator", () => {
-  test("routes preparer → planner → coder → reviewer → committer → END", () => {
+  test("routes initial → preparer → developer → submitter → END", () => {
     expect(solveIssueModerator(makeCtx(20, []))).toBe("preparer");
-    expect(solveIssueModerator(makeCtx(20, [preparerStep()]))).toBe("planner");
-    expect(solveIssueModerator(makeCtx(20, [preparerStep(), plannerStep()]))).toBe("coder");
-    expect(solveIssueModerator(makeCtx(20, [preparerStep(), plannerStep(), coderStep()]))).toBe(
-      "reviewer",
-    );
-    expect(
-      solveIssueModerator(
-        makeCtx(20, [preparerStep(), plannerStep(), coderStep(), reviewerStep(true)]),
-      ),
-    ).toBe("committer");
+    expect(solveIssueModerator(makeCtx(20, [preparerStep()]))).toBe("developer");
+    expect(solveIssueModerator(makeCtx(20, [preparerStep(), developerStep()]))).toBe("submitter");
     expect(
       solveIssueModerator(
         makeCtx(20, [
           preparerStep(),
-          plannerStep(),
-          coderStep(),
-          reviewerStep(true),
-          committerStep(),
+          developerStep(),
+          submitterStep({
+            status: "submitted",
+            prUrl: "https://github.com/example/repo/pull/1",
+          }),
         ]),
       ),
     ).toBe(END);
   });
 
-  test("reviewer rejects → coder retry when budget allows", () => {
-    const steps: ModeratorContext<SolveIssueMeta>["steps"] = [
-      plannerStep(),
-      coderStep(),
-      reviewerStep(false),
-    ];
-    expect(solveIssueModerator(makeCtx(20, steps))).toBe("coder");
-  });
-
-  test("reviewer rejects → END when max rounds exhausted", () => {
-    const steps: ModeratorContext<SolveIssueMeta>["steps"] = [
-      plannerStep(),
-      coderStep(),
-      reviewerStep(false),
-    ];
-    expect(solveIssueModerator(makeCtx(4, steps))).toBe(END);
-  });
-
-  test("multiple planner phases → coder until all complete, then reviewer", () => {
-    const phases: PlannerMeta["phases"] = [
-      {
-        hash: "AA000001",
-        title: "first phase",
-      },
-      {
-        hash: "AA000002",
-        title: "second phase",
-      },
-    ];
-    expect(solveIssueModerator(makeCtx(20, [plannerStep(phases)]))).toBe("coder");
-    expect(solveIssueModerator(makeCtx(20, [plannerStep(phases), coderStep("AA000001")]))).toBe(
-      "coder",
-    );
+  test("submitter failed → END", () => {
     expect(
       solveIssueModerator(
-        makeCtx(20, [plannerStep(phases), coderStep("AA000001"), coderStep("AA000002")]),
+        makeCtx(20, [
+          preparerStep(),
+          developerStep(),
+          submitterStep({ status: "failed", error: "gh not authenticated" }),
+        ]),
       ),
-    ).toBe("reviewer");
+    ).toBe(END);
   });
 
-  test("one-shot coder reports only last phase hash → reviewer (moderator treats as all phases done)", () => {
-    const phases: PlannerMeta["phases"] = [
-      { hash: "BB000001", title: "setup branch" },
-      { hash: "BB000002", title: "write tests" },
-      { hash: "BB000003", title: "verify" },
-      { hash: "BB000004", title: "commit and pr" },
-    ];
-    expect(solveIssueModerator(makeCtx(20, [plannerStep(phases), coderStep("BB000004")]))).toBe(
-      "reviewer",
-    );
-  });
-
-  test("unrecognised completedPhase hash → coder retry when budget allows", () => {
-    const phases: PlannerMeta["phases"] = [
-      { hash: "CC000001", title: "first phase" },
-      { hash: "CC000002", title: "second phase" },
-    ];
-    expect(solveIssueModerator(makeCtx(20, [plannerStep(phases), coderStep("all-done")]))).toBe(
-      "coder",
-    );
-  });
-
-  test("incomplete phases → END when max rounds exhausted", () => {
-    const phases: PlannerMeta["phases"] = [
-      { hash: "DD000001", title: "first phase" },
-      { hash: "DD000002", title: "second phase" },
-    ];
-    const steps: ModeratorContext<SolveIssueMeta>["steps"] = [
-      plannerStep(phases),
-      coderStep("DD000001"),
-    ];
-    expect(solveIssueModerator(makeCtx(3, steps))).toBe(END);
+  test("returns END for any unexpected last step (defensive)", () => {
+    // A submitter step with a pseudo-unknown future status would still be
+    // routed to END, since the moderator is a closed switch over known roles.
+    expect(
+      solveIssueModerator(
+        makeCtx(20, [
+          preparerStep(),
+          developerStep(),
+          submitterStep({ status: "submitted", prUrl: "https://example.com/pr/1" }),
+        ]),
+      ),
+    ).toBe(END);
   });
 });
 
@@ -296,7 +235,7 @@ describe("createSolveIssueRun", () => {
     }
   });
 
-  test("structured extraction yields preparer then planner meta from mocked chat completions", async () => {
+  test("structured extraction yields preparer meta from mocked chat completions", async () => {
     const EXPECT_PREPARER_META: PreparerMeta = {
       repoPath: "/home/user/repos/test",
       defaultBranch: "main",
@@ -308,12 +247,20 @@ describe("createSolveIssueRun", () => {
         buildCommand: "bun run build",
       },
     };
-    restoreFetch = installMockChatCompletions([EXPECT_PREPARER_META, EXPECT_PLANNER_META]);
+    restoreFetch = installMockChatCompletions([EXPECT_PREPARER_META]);
 
     casDir = await mkdtemp(join(tmpdir(), "solve-issue-cas-"));
     const cas = createCasStore(casDir);
 
-    const run = createSolveIssueRun({ agent: async () => "" }, stubExtract, null);
+    // Override developer so the test does not spin up a child workflow.
+    const run = createSolveIssueRun(
+      {
+        agent: async () => "",
+        overrides: { developer: async () => "stub-root-hash" },
+      },
+      stubExtract,
+      stubLlmProvider,
+    );
     const gen = run(
       { prompt: "task", steps: [] },
       { threadId: "01TEST000000000000000000TR", maxRounds: 20, depth: 0, cas },
@@ -325,14 +272,6 @@ describe("createSolveIssueRun", () => {
     }
     expect(first.value.role).toBe("preparer");
     expect(first.value.meta).toEqual(EXPECT_PREPARER_META);
-
-    const second = await gen.next();
-    expect(second.done).toBe(false);
-    if (second.done) {
-      throw new Error("expected yield");
-    }
-    expect(second.value.role).toBe("planner");
-    expect(second.value.meta).toEqual(EXPECT_PLANNER_META);
   });
 
   test("per-role agent overrides default", async () => {
@@ -342,11 +281,17 @@ describe("createSolveIssueRun", () => {
       conventions: null,
       toolchain: { packageManager: null, testCommand: null, lintCommand: null, buildCommand: null },
     };
-    restoreFetch = installMockChatCompletions([
-      PREPARER_META,
-      EXPECT_PLANNER_META,
-      EXPECT_CODER_META,
-    ]);
+    const DEVELOPER_META: DeveloperMeta = {
+      branch: "feat/x",
+      commitSha: "abc1234",
+      filesChanged: ["a.ts"],
+      summary: "did the work",
+    };
+    const SUBMITTER_META: SubmitterMeta = {
+      status: "submitted",
+      prUrl: "https://github.com/example/repo/pull/2",
+    };
+    restoreFetch = installMockChatCompletions([PREPARER_META, DEVELOPER_META, SUBMITTER_META]);
 
     casDir = await mkdtemp(join(tmpdir(), "solve-issue-cas-"));
     const cas = createCasStore(casDir);
@@ -363,18 +308,18 @@ describe("createSolveIssueRun", () => {
             calls.push("preparer");
             return "";
           },
-          planner: async () => {
-            calls.push("planner");
-            return "";
+          developer: async () => {
+            calls.push("developer");
+            return "stub-root-hash";
           },
-          coder: async () => {
-            calls.push("coder");
+          submitter: async () => {
+            calls.push("submitter");
             return "";
           },
         },
       },
       stubExtract,
-      null,
+      stubLlmProvider,
     );
     const gen = run(
       { prompt: "task", steps: [] },
@@ -385,16 +330,65 @@ describe("createSolveIssueRun", () => {
 
     calls.length = 0;
     await gen.next();
-    expect(calls).toEqual(["planner"]);
+    expect(calls).toEqual(["developer"]);
 
     calls.length = 0;
     await gen.next();
-    expect(calls).toEqual(["coder"]);
+    expect(calls).toEqual(["submitter"]);
+  });
+
+  test("developer defaults to workflowAsAgent override (caller override still wins)", async () => {
+    const PREPARER_META: PreparerMeta = {
+      repoPath: "/tmp/r",
+      defaultBranch: "main",
+      conventions: null,
+      toolchain: { packageManager: null, testCommand: null, lintCommand: null, buildCommand: null },
+    };
+    const DEVELOPER_META: DeveloperMeta = {
+      branch: "feat/y",
+      commitSha: "def5678",
+      filesChanged: ["b.ts"],
+      summary: "more work",
+    };
+    restoreFetch = installMockChatCompletions([PREPARER_META, DEVELOPER_META]);
+
+    casDir = await mkdtemp(join(tmpdir(), "solve-issue-cas-"));
+    const cas = createCasStore(casDir);
+
+    let developerInvocations = 0;
+    const run = createSolveIssueRun(
+      {
+        agent: async () => "",
+        overrides: {
+          developer: async () => {
+            developerInvocations += 1;
+            return "stub-root-hash";
+          },
+        },
+      },
+      stubExtract,
+      stubLlmProvider,
+    );
+    const gen = run(
+      { prompt: "task", steps: [] },
+      { threadId: "01TEST000000000000000000TR", maxRounds: 20, depth: 0, cas },
+    );
+    // preparer
+    await gen.next();
+    // developer (caller override should be invoked, NOT workflowAsAgent default)
+    const devYield = await gen.next();
+    expect(devYield.done).toBe(false);
+    if (devYield.done) {
+      throw new Error("expected yield");
+    }
+    expect(devYield.value.role).toBe("developer");
+    expect(devYield.value.meta).toEqual(DEVELOPER_META);
+    expect(developerInvocations).toBe(1);
   });
 });
 
 describe("buildSolveIssueDescriptor", () => {
-  test("lists all roles with schemas that validate", () => {
+  test("lists preparer, developer, submitter with schemas that validate", () => {
     const descriptor = buildSolveIssueDescriptor();
     const validated = validateWorkflowDescriptor(descriptor);
     expect(validated.ok).toBe(true);
@@ -402,13 +396,11 @@ describe("buildSolveIssueDescriptor", () => {
       throw new Error(validated.error);
     }
     expect(Object.keys(validated.value.roles).sort()).toEqual([
-      "coder",
-      "committer",
-      "planner",
+      "developer",
       "preparer",
-      "reviewer",
+      "submitter",
     ]);
-    for (const key of ["preparer", "planner", "coder", "reviewer", "committer"] as const) {
+    for (const key of ["preparer", "developer", "submitter"] as const) {
       const role = validated.value.roles[key];
       expect(role).toBeDefined();
       expect(typeof role.schema).toBe("object");

packages/workflow-template-solve-issue/package.json

@@ -10,10 +10,8 @@
   },
   "dependencies": {
     "@uncaged/workflow": "workspace:*",
-    "@uncaged/workflow-role-committer": "workspace:*",
-    "@uncaged/workflow-role-coder": "workspace:*",
-    "@uncaged/workflow-role-planner": "workspace:*",
     "@uncaged/workflow-role-preparer": "workspace:*",
-    "@uncaged/workflow-role-reviewer": "workspace:*"
+    "@uncaged/workflow-role-submitter": "workspace:*",
+    "zod": "^4.0.0"
   }
 }

packages/workflow-template-solve-issue/src/developer.ts

@@ -0,0 +1,37 @@
+import type { RoleDefinition } from "@uncaged/workflow";
+import * as z from "zod/v4";
+
+export const developerMetaSchema = z.object({
+  branch: z.string(),
+  commitSha: z.string(),
+  filesChanged: z.array(z.string()),
+  summary: z.string(),
+});
+
+export type DeveloperMeta = z.infer<typeof developerMetaSchema>;
+
+const DEVELOPER_SYSTEM = `You are the **developer**. You delegate the implementation work to the \`develop\` workflow.
+
+The actual implementation (planning → coding → reviewing → testing → committing) is handled by a child workflow that runs in your place. Your output is the Merkle DAG root hash of that child thread.
+
+Pass through the task and let the child workflow do the work.`;
+
+const DEVELOPER_EXTRACT_PROMPT = `The agent output is the root CAS hash of a child workflow thread. Use the cas_get tool to traverse the Merkle DAG and extract the developer summary.
+
+Procedure:
+1. cas_get(<rootHash>) — the root node lists all child step hashes (planner, coder, reviewer, tester, committer).
+2. Find the committer step. cas_get its hash to read the committer's meta — extract branch and commitSha from there.
+3. Find every coder step. cas_get each to read the coder's filesChanged. Union all filesChanged across coder steps.
+4. Compose a short human-readable summary describing what the develop child workflow accomplished (drawn from the coder summaries, or a synthesis of them).
+
+Return: { branch, commitSha, filesChanged, summary }.`;
+
+export const developerRole: RoleDefinition<DeveloperMeta> = {
+  description:
+    "Delegates the actual implementation to the develop workflow (workflow-as-agent). Produces a summary by traversing the child thread's Merkle DAG.",
+  systemPrompt: DEVELOPER_SYSTEM,
+  extractPrompt: DEVELOPER_EXTRACT_PROMPT,
+  schema: developerMetaSchema,
+  extractRefs: () => [],
+  extractMode: "react",
+};

packages/workflow-template-solve-issue/src/index.ts

@@ -5,38 +5,28 @@ import {
   type LlmProvider,
   type WorkflowDefinition,
   type WorkflowFn,
+  workflowAsAgent,
 } from "@uncaged/workflow";
 
 import { solveIssueModerator } from "./moderator.js";
 import { SOLVE_ISSUE_WORKFLOW_DESCRIPTION, type SolveIssueMeta, solveIssueRoles } from "./roles.js";
 
-export {
-  type CoderMeta,
-  coderMetaSchema,
-  coderRole,
-} from "@uncaged/workflow-role-coder";
-export {
-  type CommitterMeta,
-  committerMetaSchema,
-  committerRole,
-} from "@uncaged/workflow-role-committer";
-export {
-  type PlannerMeta,
-  phaseSchema,
-  plannerMetaSchema,
-  plannerRole,
-} from "@uncaged/workflow-role-planner";
 export {
   type PreparerMeta,
   preparerMetaSchema,
   preparerRole,
 } from "@uncaged/workflow-role-preparer";
 export {
-  type ReviewerMeta,
-  reviewerMetaSchema,
-  reviewerRole,
-} from "@uncaged/workflow-role-reviewer";
+  type SubmitterMeta,
+  submitterMetaSchema,
+  submitterRole,
+} from "@uncaged/workflow-role-submitter";
 export { buildSolveIssueDescriptor } from "./descriptor.js";
+export {
+  type DeveloperMeta,
+  developerMetaSchema,
+  developerRole,
+} from "./developer.js";
 export { solveIssueModerator } from "./moderator.js";
 export {
   SOLVE_ISSUE_WORKFLOW_DESCRIPTION,
@@ -51,10 +41,25 @@ export const solveIssueWorkflowDefinition: WorkflowDefinition<SolveIssueMeta> =
   moderator: solveIssueModerator,
 };
 
+/**
+ * Build the solve-issue {@link WorkflowFn}.
+ *
+ * The `developer` role always delegates to the registered `develop` workflow via
+ * {@link workflowAsAgent}; if the caller supplies their own `developer` override in
+ * `binding.overrides`, it takes precedence so tests and custom hosts can stub it.
+ */
 export function createSolveIssueRun(
   binding: AgentBinding,
   extract: ExtractFn,
   llmProvider: LlmProvider | null,
 ): WorkflowFn {
-  return createWorkflow(solveIssueWorkflowDefinition, binding, extract, llmProvider);
+  const developerOverride = binding.overrides?.developer ?? workflowAsAgent("develop");
+  const mergedBinding: AgentBinding = {
+    agent: binding.agent,
+    overrides: {
+      ...(binding.overrides ?? {}),
+      developer: developerOverride,
+    },
+  };
+  return createWorkflow(solveIssueWorkflowDefinition, mergedBinding, extract, llmProvider);
 }

packages/workflow-template-solve-issue/src/moderator.ts

@@ -1,52 +1,9 @@
-import type { Moderator, ModeratorContext } from "@uncaged/workflow";
+import type { Moderator } from "@uncaged/workflow";
 import { END } from "@uncaged/workflow";
 
 import type { SolveIssueMeta } from "./roles.js";
 
-function coderFinishedAllPlannedPhases(
-  phases: ReadonlyArray<{ hash: string }>,
-  coderCompletedPhases: ReadonlyArray<string>,
-): boolean {
-  if (phases.length === 0) {
-    return true;
-  }
-  const plannedHashes = new Set(phases.map((p) => p.hash));
-  const lastHash = phases[phases.length - 1].hash;
-  const explicit = new Set(coderCompletedPhases.filter((h) => plannedHashes.has(h)));
-  if (phases.every((p) => explicit.has(p.hash))) {
-    return true;
-  }
-  if (coderCompletedPhases.some((h) => h === lastHash)) {
-    return true;
-  }
-  return false;
-}
-
-function nextAfterCoder(
-  ctx: ModeratorContext<SolveIssueMeta>,
-  maxRounds: number,
-): (keyof SolveIssueMeta & string) | typeof END {
-  const plannerStep = ctx.steps.find((s) => s.role === "planner");
-  if (plannerStep === undefined) {
-    return "reviewer";
-  }
-  const phases = plannerStep.meta.phases;
-  const coderCompletedPhases = ctx.steps
-    .filter((s) => s.role === "coder")
-    .map((s) => s.meta.completedPhase);
-  const allDone = coderFinishedAllPlannedPhases(phases, coderCompletedPhases);
-  if (allDone) {
-    return "reviewer";
-  }
-  if (ctx.steps.length < maxRounds - 1) {
-    return "coder";
-  }
-  return END;
-}
-
 export const solveIssueModerator: Moderator<SolveIssueMeta> = (ctx) => {
-  const maxRounds = ctx.start.meta.maxRounds;
-
   if (ctx.steps.length === 0) {
     return "preparer";
   }
@@ -54,31 +11,14 @@ export const solveIssueModerator: Moderator<SolveIssueMeta> = (ctx) => {
   const last = ctx.steps[ctx.steps.length - 1];
 
   if (last.role === "preparer") {
-    return "planner";
+    return "developer";
   }
 
-  if (last.role === "planner") {
-    return "coder";
+  if (last.role === "developer") {
+    return "submitter";
   }
 
-  if (last.role === "coder") {
-    return nextAfterCoder(ctx, maxRounds);
-  }
-
-  if (last.role === "reviewer") {
-    if (last.meta.status === "approved") {
-      return "committer";
-    }
-    if (ctx.steps.length < maxRounds - 1) {
-      return "coder";
-    }
-    return END;
-  }
-
-  if (last.role === "committer") {
-    if (last.meta.status === "recoverable" && ctx.steps.length < maxRounds - 1) {
-      return "coder";
-    }
+  if (last.role === "submitter") {
     return END;
   }
 

packages/workflow-template-solve-issue/src/roles.ts

@@ -1,19 +1,16 @@
 import type { RoleDefinition } from "@uncaged/workflow";
-import { type CoderMeta, coderRole } from "@uncaged/workflow-role-coder";
-import { type CommitterMeta, committerRole } from "@uncaged/workflow-role-committer";
-import { type PlannerMeta, plannerRole } from "@uncaged/workflow-role-planner";
 import { type PreparerMeta, preparerRole } from "@uncaged/workflow-role-preparer";
-import { type ReviewerMeta, reviewerRole } from "@uncaged/workflow-role-reviewer";
+import { type SubmitterMeta, submitterRole } from "@uncaged/workflow-role-submitter";
+
+import { type DeveloperMeta, developerRole } from "./developer.js";
 
 export const SOLVE_ISSUE_WORKFLOW_DESCRIPTION =
-  "Prepare repo context, plan phases, implement incrementally, review, and commit to resolve an issue end-to-end (preparer → planner → coder [repeat per phase] → reviewer → committer).";
+  "Resolve an issue end-to-end by preparing the repo, delegating implementation to the develop workflow, and opening a pull request (preparer → developer → submitter).";
 
 export type SolveIssueMeta = {
   preparer: PreparerMeta;
-  planner: PlannerMeta;
-  coder: CoderMeta;
-  reviewer: ReviewerMeta;
-  committer: CommitterMeta;
+  developer: DeveloperMeta;
+  submitter: SubmitterMeta;
 };
 
 export type SolveIssueRoles = {
@@ -22,8 +19,6 @@ export type SolveIssueRoles = {
 
 export const solveIssueRoles: SolveIssueRoles = {
   preparer: preparerRole,
-  planner: plannerRole,
-  coder: coderRole,
-  reviewer: reviewerRole,
-  committer: committerRole,
+  developer: developerRole,
+  submitter: submitterRole,
 };

packages/workflow-template-solve-issue/tsconfig.json

@@ -8,9 +8,7 @@
   "include": ["src/**/*.ts"],
   "references": [
     { "path": "../workflow" },
-    { "path": "../workflow-role-coder" },
-    { "path": "../workflow-role-committer" },
-    { "path": "../workflow-role-planner" },
-    { "path": "../workflow-role-reviewer" }
+    { "path": "../workflow-role-preparer" },
+    { "path": "../workflow-role-submitter" }
   ]
 }

packages/workflow/__tests__/fork-thread.test.ts

@@ -87,6 +87,26 @@ describe("fork-thread", () => {
     expect(r.value.runOptions).toEqual({ maxRounds: 5, depth: 0 });
   });
 
+  test("parseThreadDataJsonl ignores trailing WorkflowResult line", () => {
+    const text = `${sampleDataJsonl.trim()}\n{"returnCode":0,"summary":"done"}\n`;
+    const r = parseThreadDataJsonl(text);
+    expect(r.ok).toBe(true);
+    if (!r.ok) {
+      return;
+    }
+    expect(r.value.roleSteps.length).toBe(3);
+    expect(r.value.roleSteps[2]?.role).toBe("reviewer");
+  });
+
+  test("parseThreadDataJsonl errors when WorkflowResult is not last", () => {
+    const text = `{"name":"demo","hash":"H","threadId":"01ZZZZZZZZZZZZZZZZZZZZZZ","parameters":{"prompt":"p","options":{"maxRounds":3}},"timestamp":1}
+{"returnCode":0,"summary":"early"}
+{"role":"planner","content":"x","meta":{},"timestamp":2}
+`;
+    const r = parseThreadDataJsonl(text);
+    expect(r.ok).toBe(false);
+  });
+
   test("parseThreadDataJsonl reads explicit depth from start record", () => {
     const text = `{"name":"demo","hash":"H","threadId":"01ZZZZZZZZZZZZZZZZZZZZZZ","parameters":{"prompt":"p","options":{"maxRounds":3,"depth":2}},"timestamp":1}
 {"role":"planner","contentHash":"HP0000000000000000000099","meta":{},"refs":[],"timestamp":2}

packages/workflow/__tests__/worker.test.ts

@@ -125,7 +125,7 @@ describe("worker process", () => {
           .trim()
           .split("\n")
           .filter((l) => l !== "").length,
-      ).toBe(3);
+      ).toBe(4);
     } finally {
       await rm(root, { recursive: true, force: true });
     }
@@ -187,7 +187,7 @@ describe("worker process", () => {
         .trim()
         .split("\n")
         .filter((l) => l !== "");
-      expect(lines.length).toBe(3);
+      expect(lines.length).toBe(4);
       const start = JSON.parse(lines[0] ?? "{}") as Record<string, unknown>;
       expect(start.forkFrom).toEqual({ threadId: srcId });
       const replay = JSON.parse(lines[1] ?? "{}") as Record<string, unknown>;
@@ -195,6 +195,8 @@ describe("worker process", () => {
       expect(replay.timestamp).toBe(555);
       const coder = JSON.parse(lines[2] ?? "{}") as Record<string, unknown>;
       expect(coder.role).toBe("coder");
+      const done = JSON.parse(lines[3] ?? "{}") as Record<string, unknown>;
+      expect(done.returnCode).toBe(0);
     } finally {
       await rm(root, { recursive: true, force: true });
     }

packages/workflow/src/fork-thread.ts

@@ -1,6 +1,6 @@
 import { normalizeRefsField } from "./refs-field.js";
 import { err, ok, type Result } from "./result.js";
-import type { RoleOutput } from "./types.js";
+import type { RoleOutput, WorkflowCompletion } from "./types.js";
 
 /** Role steps replayed from `.data.jsonl`, including persisted timestamps. */
 export type ForkHistoricalStep = RoleOutput & { timestamp: number };
@@ -14,33 +14,56 @@ export type ParsedThreadStartRecord = {
   depth: number;
 };
 
-function parseRoleLine(
+/** Recognizes a persisted workflow completion line (no `role`; has numeric `returnCode` and string `summary`). Omits `rootHash` when absent. */
+export function tryParseWorkflowResultRecord(
   obj: Record<string, unknown>,
-  lineIndex: number,
-): Result<ForkHistoricalStep, string> {
+): WorkflowCompletion | null {
+  if (obj.role !== undefined) {
+    return null;
+  }
+  const returnCode = obj.returnCode;
+  const summary = obj.summary;
+  if (typeof returnCode !== "number" || typeof summary !== "string") {
+    return null;
+  }
+  return { returnCode, summary };
+}
+
+export function tryParseRoleStepRecord(obj: Record<string, unknown>): ForkHistoricalStep | null {
   const role = obj.role;
   const contentHash = obj.contentHash;
   const meta = obj.meta;
   const timestamp = obj.timestamp;
   if (typeof role !== "string") {
-    return err(`invalid role record at line ${lineIndex}: missing role`);
+    return null;
   }
   if (typeof contentHash !== "string") {
-    return err(`invalid role record at line ${lineIndex}: missing contentHash`);
+    return null;
   }
   if (meta === null || typeof meta !== "object") {
-    return err(`invalid role record at line ${lineIndex}: missing meta`);
+    return null;
   }
   if (typeof timestamp !== "number") {
-    return err(`invalid role record at line ${lineIndex}: missing timestamp`);
+    return null;
   }
-  return ok({
+  return {
     role,
     contentHash,
     meta: meta as Record<string, unknown>,
     refs: normalizeRefsField(obj.refs),
     timestamp,
-  });
+  };
+}
+
+function parseRoleLine(
+  obj: Record<string, unknown>,
+  lineIndex: number,
+): Result<ForkHistoricalStep, string> {
+  const parsed = tryParseRoleStepRecord(obj);
+  if (parsed === null) {
+    return err(`invalid role record at line ${lineIndex}`);
+  }
+  return ok(parsed);
 }
 
 function parseStartRecordLine(firstLine: string): Result<ParsedThreadStartRecord, string> {
@@ -109,7 +132,15 @@ function parseFollowingRoleLines(lines: string[]): Result<ForkHistoricalStep[],
     if (rec === null || typeof rec !== "object") {
       return err(`invalid record at line ${i + 1}`);
     }
-    const parsed = parseRoleLine(rec as Record<string, unknown>, i + 1);
+    const recObj = rec as Record<string, unknown>;
+    const wf = tryParseWorkflowResultRecord(recObj);
+    if (wf !== null) {
+      if (i !== lines.length - 1) {
+        return err("WorkflowResult record must be the final line in `.data.jsonl`");
+      }
+      break;
+    }
+    const parsed = parseRoleLine(recObj, i + 1);
     if (!parsed.ok) {
       return parsed;
     }

packages/workflow/src/index.ts

@@ -25,6 +25,8 @@ export {
   type ParsedThreadStartRecord,
   parseThreadDataJsonl,
   selectForkHistoricalSteps,
+  tryParseRoleStepRecord,
+  tryParseWorkflowResultRecord,
 } from "./fork-thread.js";
 export { type GcResult, garbageCollectCas } from "./gc.js";
 export { stringifyWorkflowDescriptor } from "./generate-descriptor.js";

packages/workflow/src/worker.ts

@@ -1,4 +1,4 @@
-import { mkdir, unlink, writeFile } from "node:fs/promises";
+import { appendFile, mkdir, unlink, writeFile } from "node:fs/promises";
 import { createServer, type Socket } from "node:net";
 import { dirname, join } from "node:path";
 import { importWorkflowBundleModule } from "./bundle-import-env.js";
@@ -11,7 +11,7 @@ import { normalizeRefsField } from "./refs-field.js";
 import { err, ok, type Result } from "./result.js";
 import { getGlobalCasDir } from "./storage-root.js";
 import { createThreadPauseGate, type ThreadPauseGate } from "./thread-pause-gate.js";
-import type { RoleOutput, WorkflowFn } from "./types.js";
+import type { RoleOutput, WorkflowFn, WorkflowResult } from "./types.js";
 
 const bootLog = createLogger({ sink: { kind: "stderr" } });
 
@@ -404,7 +404,7 @@ async function main(): Promise<void> {
         });
       }
 
-      await executeThread(
+      const runResult = await executeThread(
         workflowFn,
         cmd.workflowName,
         { prompt: cmd.prompt, steps: cmd.steps },
@@ -418,9 +418,12 @@ async function main(): Promise<void> {
         io,
         logger,
       );
+      await appendFile(dataJsonlPath, `${JSON.stringify(runResult)}\n`, "utf8");
     } catch (e) {
       const message = e instanceof Error ? e.message : String(e);
       bootLog("Q3MN8YKW", `thread ${threadId} failed: ${message}`);
+      const failure: WorkflowResult = { returnCode: 1, summary: message, rootHash: "" };
+      await appendFile(dataJsonlPath, `${JSON.stringify(failure)}\n`, "utf8").catch(() => {});
     } finally {
       threads.delete(threadId);
       await unlink(runningPath).catch(() => {});

packages/workflow/workflow

@@ -0,0 +1 @@
+/home/azureuser/repos/uncaged-workflow/packages/workflow

tsconfig.json

@@ -22,11 +22,15 @@
     { "path": "packages/workflow-role-committer" },
     { "path": "packages/workflow-role-coder" },
     { "path": "packages/workflow-role-planner" },
+    { "path": "packages/workflow-role-preparer" },
     { "path": "packages/workflow-role-reviewer" },
+    { "path": "packages/workflow-role-submitter" },
+    { "path": "packages/workflow-role-tester" },
     { "path": "packages/workflow-agent-cursor" },
     { "path": "packages/workflow-agent-hermes" },
     { "path": "packages/workflow-util-agent" },
     { "path": "packages/cli-workflow" },
-    { "path": "packages/workflow-template-solve-issue" }
+    { "path": "packages/workflow-template-solve-issue" },
+    { "path": "packages/workflow-template-develop" }
   ]
 }