fix: setup UX improvements — adapter check, ENOENT, SQLite warning, VERSION, PATH docs

- setup validates adapter binary availability, prints install command if missing
- setup prints 'Config saved to <path> ✓' on success
- spawn ENOENT gives actionable error with which command
- SQLite ExperimentalWarning suppressed via NODE_OPTIONS
- bootstrap VERSION reads cli package.json (was reading util)
- bootstrap PATH guidance is shell-agnostic

Fixes #114

.changeset/bootstrap-onboarding.md

@@ -0,0 +1,9 @@
+---
+"@united-workforce/cli": patch
+---
+
+fix: expand bootstrap prompt with full onboarding and upgrade guide
+
+Bootstrap now covers two scenarios:
+- Fresh install: CLI + adapter installation, `uwf setup` configuration, skill installation, end-to-end verification
+- Upgrade: package update, skill regeneration, breaking change migrations (e.g. $START new/resume)

.changeset/bootstrap-preflight.md

@@ -0,0 +1,8 @@
+---
+"@united-workforce/cli": patch
+---
+
+fix: bootstrap adds Step 0 environment pre-flight check
+
+- Pre-flight checks for node, pnpm/npm, global bin PATH, hermes CLI with FIX instructions (#112)
+- Install commands changed from npm to pnpm (with npm fallback)

.changeset/bootstrap-schema-path.md

@@ -0,0 +1,9 @@
+---
+"@united-workforce/cli": patch
+"@united-workforce/util": patch
+---
+
+fix: workflow-authoring flat schema example uses enum, bootstrap adds PATH guidance
+
+- workflow-authoring: flat schema example uses `enum: [done]` instead of bare `const` (#110.3)
+- bootstrap: adds `which hermes` check and PATH guidance for venv installs (#110.4)

.changeset/bootstrap-ux-fixes.md

@@ -0,0 +1,10 @@
+---
+"@united-workforce/cli": patch
+---
+
+fix: preset provider base-url auto-fill, bootstrap ACP docs, friendlier name mismatch error
+
+- `uwf setup --provider dashscope` now auto-fills `--base-url` from preset list (#106)
+- Bootstrap guide documents uwf-hermes ACP dependency (`pip install hermes-agent[acp]`) (#107)
+- Bootstrap verify step uses inline workflow instead of missing `examples/eval-simple.yaml` (#107)
+- Workflow filename mismatch error now suggests how to fix it (#108)

.changeset/setup-ux-improvements.md

@@ -0,0 +1,12 @@
+---
+"@united-workforce/cli": patch
+---
+
+fix: setup UX improvements (#114)
+
+- Setup validates adapter availability and prints install command if missing
+- Setup prints "Config saved to <path> ✓" on success
+- Spawn ENOENT gives actionable error ("not found in PATH" + which command)
+- SQLite ExperimentalWarning suppressed via NODE_OPTIONS in spawned processes
+- Bootstrap VERSION reads cli package version (was reading util version)
+- Bootstrap PATH guidance is shell-agnostic (no hardcoded .bashrc/.profile)

.changeset/start-new-resume.md

@@ -0,0 +1,9 @@
+---
+"@united-workforce/cli": minor
+"@united-workforce/util": patch
+---
+
+feat: replace $START `_` status with `new`/`resume` semantics
+
+BREAKING: All workflow YAML files must update `$START._` to `$START.new` + `$START.resume`.
+The `resume` edge prompt replaces the previously hardcoded resume message in the CLI.

.workflows/e2e-walkthrough.yaml

@@ -264,7 +264,8 @@ roles:
 
 graph:
   $START:
-    _: { role: "bootstrap", prompt: "Set up the Docker container and verify uwf is runnable." }
+    new: { role: "bootstrap", prompt: "Set up the Docker container and verify uwf is runnable." }
+    resume: { role: "bootstrap", prompt: "Review the previous run output and continue the walkthrough." }
   bootstrap:
     pass: { role: "config-and-registry", prompt: "Container {{{containerName}}} is ready. Validate config and workflow registration." }
     fail: { role: "$END", prompt: "Bootstrap failed: {{{error}}}. No container was created." }

.workflows/solve-issue.yaml

@@ -227,7 +227,8 @@ roles:
           required: [$status, error]
 graph:
   $START:
-    _: { role: "planner", prompt: "Analyze the issue and produce an implementation plan." }
+    new: { role: "planner", prompt: "Analyze the issue and produce an implementation plan." }
+    resume: { role: "planner", prompt: "Review the previous run output and continue the work." }
   planner:
     insufficient_info: { role: "$SUSPEND", prompt: "信息不足，需要补充：{{{reason}}}" }
     ready: { role: "developer", prompt: "Implement the TDD test spec (CAS hash: {{{plan}}}) in repo {{{repoPath}}}. Repo remote: {{{repoRemote}}}." }

docs/wf-stateless-design.md

@@ -200,7 +200,7 @@ payload:
 
 - `roles` — 内联定义，每个 role 的 `meta` 是独立的 ocas_ref（指向 ocas 内置 JSON Schema 节点）
 - `graph` — `Record<Role | "$START", Record<Status, Target>>`，每个 Target = `{ role, prompt }`
-- Status 来自上一个 role 输出的 `status` 字段，`$START` 用 `_` 作为初始 status
+- Status 来自上一个 role 输出的 `$status` 字段，`$START` 使用 `new`（首次启动）和 `resume`（恢复已完成的 thread）作为 status
 - Prompt 模板使用 Mustache 渲染，变量来自 lastOutput
 - 不含 agent binding — agent 配置在 `~/.uwf/config.yaml` 中管理
 
@@ -208,7 +208,7 @@ Moderator 的求值逻辑：
 
 ```typescript
 evaluate(graph, lastRole, lastOutput) → { role, prompt }
-// 1. status = lastRole === "$START" ? "_" : lastOutput.status
+// 1. status = lastOutput.$status (e.g. "new" for $START first run, "resume" for completed thread resume)
 // 2. target = graph[lastRole][status]
 // 3. prompt = mustache.render(target.prompt, lastOutput)
 ```
@@ -422,8 +422,8 @@ type StepNodePayload = StepRecord & {
 Moderator 使用 `evaluate(graph, lastRole, lastOutput)` 进行同步 status-based routing：
 
 ```typescript
-// graph[lastRole][lastOutput.status] → Target { role, prompt }
+// graph[lastRole][lastOutput.$status] → Target { role, prompt }
-// $START 角色使用 "_" 作为初始 status
+// $START 使用 "new"（首次启动）和 "resume"（恢复已完成 thread）作为 status
 // prompt 通过 Mustache 模板渲染，变量来自 lastOutput
 ```
 

examples/analyze-topic.yaml

@@ -35,6 +35,7 @@ roles:
       required: [$status, thesis, keyPoints]
 graph:
   $START:
-    _: { role: "analyst", prompt: "Analyze the topic in the task and produce a structured summary with key points." }
+    new: { role: "analyst", prompt: "Analyze the topic in the task and produce a structured summary with key points." }
+    resume: { role: "analyst", prompt: "Review the previous analysis output and continue with additional context." }
   analyst:
     done: { role: "$END", prompt: "Analysis complete. Finish the workflow." }

examples/debate.yaml

@@ -53,7 +53,8 @@ roles:
       required: [$status, argument]
 graph:
   $START:
-    _: { role: "against", prompt: "Present your opening argument against the proposition." }
+    new: { role: "against", prompt: "Present your opening argument against the proposition." }
+    resume: { role: "against", prompt: "Review the previous debate output and continue the argument against the proposition." }
   against:
     conceded: { role: "$END", prompt: "The against side conceded. Debate over." }
     continue: { role: "for", prompt: "Counter the opposing argument: {{{argument}}}" }

examples/eval-simple.yaml

@@ -25,6 +25,7 @@ roles:
       required: [$status, summary]
 graph:
   $START:
-    _: { role: "fixer", prompt: "Fix the code issue described in the task prompt." }
+    new: { role: "fixer", prompt: "Fix the code issue described in the task prompt." }
+    resume: { role: "fixer", prompt: "Review the previous run output and continue fixing the code issue." }
   fixer:
     done: { role: "$END", prompt: "Fix complete." }

examples/solve-issue.yaml

@@ -215,7 +215,8 @@ roles:
           required: [$status, error]
 graph:
   $START:
-    _: { role: "planner", prompt: "Analyze the issue and produce an implementation plan." }
+    new: { role: "planner", prompt: "Analyze the issue and produce an implementation plan." }
+    resume: { role: "planner", prompt: "Review the previous run output and continue the work." }
   planner:
     insufficient_info: { role: "$SUSPEND", prompt: "信息不足，需要补充：{{{reason}}}" }
     ready: { role: "developer", prompt: "Implement the TDD test spec (CAS hash: {{{plan}}}) in repo {{{repoPath}}}." }

packages/agent-builtin/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/agent-builtin",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "files": [
     "src",
     "dist",

packages/agent-builtin/src/cli.ts

@@ -1,5 +1,12 @@
 #!/usr/bin/env node
 
+// eslint-disable-next-line -- dynamic import for version
+const pkg = await import("../package.json", { with: { type: "json" } });
+if (process.argv.includes("--version") || process.argv.includes("-V")) {
+  process.stdout.write(`${pkg.default.version}\n`);
+  process.exit(0);
+}
+
 import { createBuiltinAgent } from "./agent.js";
 
 const main = createBuiltinAgent();

packages/agent-claude-code/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/agent-claude-code",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "files": [
     "src",
     "dist",

packages/agent-claude-code/src/cli.ts

@@ -1,5 +1,12 @@
 #!/usr/bin/env node
 
+// eslint-disable-next-line -- dynamic import for version
+const pkg = await import("../package.json", { with: { type: "json" } });
+if (process.argv.includes("--version") || process.argv.includes("-V")) {
+  process.stdout.write(`${pkg.default.version}\n`);
+  process.exit(0);
+}
+
 import { createClaudeCodeAgent } from "./claude-code.js";
 
 const model = process.env.CLAUDE_MODEL ?? null;

packages/agent-hermes/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/agent-hermes",
-  "version": "0.1.1",
+  "version": "0.1.2",
   "files": [
     "src",
     "dist",

packages/agent-hermes/src/acp-client.ts

@@ -1,6 +1,16 @@
 import type { ChildProcess } from "node:child_process";
 import { spawn } from "node:child_process";
+import { readFileSync } from "node:fs";
+import { dirname, join } from "node:path";
 import { createInterface } from "node:readline";
+import { fileURLToPath } from "node:url";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const OWN_VERSION = (
+  JSON.parse(readFileSync(join(__dirname, "..", "package.json"), "utf-8")) as {
+    version: string;
+  }
+).version;
 
 const HERMES_COMMAND = "hermes";
 const PROTOCOL_VERSION = 1;
@@ -299,7 +309,7 @@ export class HermesAcpClient {
   private async initialize(): Promise<void> {
     const initResponse = await this.sendRequest("initialize", {
       protocolVersion: PROTOCOL_VERSION,
-      clientInfo: { name: "uwf", version: "0.1.0" },
+      clientInfo: { name: "uwf-hermes", version: OWN_VERSION },
       capabilities: {},
     });
 

packages/agent-hermes/src/cli.ts

@@ -1,5 +1,12 @@
 #!/usr/bin/env node
 
+// eslint-disable-next-line -- dynamic import for version
+const pkg = await import("../package.json", { with: { type: "json" } });
+if (process.argv.includes("--version") || process.argv.includes("-V")) {
+  process.stdout.write(`${pkg.default.version}\n`);
+  process.exit(0);
+}
+
 import { createHermesAgent } from "./hermes.js";
 import { isResumeDisabled } from "./session-cache.js";
 

packages/agent-mock/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/agent-mock",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "files": [
     "src",
     "dist",

packages/agent-mock/src/cli.ts

@@ -1,5 +1,12 @@
 #!/usr/bin/env node
 
+// eslint-disable-next-line -- dynamic import for version
+const pkg = await import("../package.json", { with: { type: "json" } });
+if (process.argv.includes("--version") || process.argv.includes("-V")) {
+  process.stdout.write(`${pkg.default.version}\n`);
+  process.exit(0);
+}
+
 import { createMockAgent } from "./mock-agent.js";
 
 const USAGE = "usage: uwf-mock --mock-data <path> --thread <id> --role <role> --prompt <text>";

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/cli",
-  "version": "0.1.1",
+  "version": "0.2.0",
   "files": [
     "src",
     "dist",

packages/cli/src/__tests__/adapter-json-roundtrip.test.ts

@@ -58,7 +58,10 @@ describe("C1: adapter JSON round-trip integration", () => {
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "Do the work", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "Do the work", location: null },
+          resume: { role: "worker", prompt: "Resume the work", location: null },
+        },
         worker: { done: { role: "$END", prompt: "completed", location: null } },
       },
     });

packages/cli/src/__tests__/current-role.test.ts

@@ -45,10 +45,14 @@ roles:
         $status: { type: string, enum: ["done"] }
 graph:
   $START:
-    _:
+    new:
       role: roleA
       prompt: "Do A"
       location: null
+    resume:
+      role: roleA
+      prompt: "Resume A"
+      location: null
   roleA:
     ready:
       role: roleB
@@ -107,10 +111,14 @@ roles:
         $status: { type: string, enum: ["done"] }
 graph:
   $START:
-    _:
+    new:
       role: roleA
       prompt: "Do A"
       location: null
+    resume:
+      role: roleA
+      prompt: "Resume A"
+      location: null
   roleA:
     pass:
       role: roleB
@@ -150,10 +158,14 @@ roles:
         $status: { type: string, enum: ["done"] }
 graph:
   $START:
-    _:
+    new:
       role: worker
       prompt: "Work"
       location: null
+    resume:
+      role: worker
+      prompt: "Resume work"
+      location: null
   worker:
     done:
       role: $END

packages/cli/src/__tests__/fixtures/e2e-count.workflow.yaml

@@ -36,7 +36,8 @@ roles:
           required: [$status]
 graph:
   $START:
-    _: { role: analyst, prompt: 'Analyze the task' }
+    new: { role: analyst, prompt: 'Analyze the task' }
+    resume: { role: analyst, prompt: 'Review the previous run output and continue the work.' }
   analyst:
     analyzed: { role: developer, prompt: 'Implement the change' }
   developer:

packages/cli/src/__tests__/fixtures/e2e-linear.workflow.yaml

@@ -25,7 +25,8 @@ roles:
           required: [$status]
 graph:
   $START:
-    _: { role: planner, prompt: 'Plan the task' }
+    new: { role: planner, prompt: 'Plan the task' }
+    resume: { role: planner, prompt: 'Review the previous run output and continue the work.' }
   planner:
     ready: { role: worker, prompt: 'Do the work' }
   worker:

packages/cli/src/__tests__/fixtures/e2e-loop.workflow.yaml

@@ -28,7 +28,8 @@ roles:
           required: [$status]
 graph:
   $START:
-    _: { role: developer, prompt: 'Implement the change' }
+    new: { role: developer, prompt: 'Implement the change' }
+    resume: { role: developer, prompt: 'Review the previous run output and continue the work.' }
   developer:
     review_needed: { role: reviewer, prompt: 'Review the change' }
   reviewer:

packages/cli/src/__tests__/fixtures/e2e-mustache.workflow.yaml

@@ -27,7 +27,8 @@ roles:
           required: [$status]
 graph:
   $START:
-    _: { role: planner, prompt: 'Plan the task' }
+    new: { role: planner, prompt: 'Plan the task' }
+    resume: { role: planner, prompt: 'Review the previous run output and continue the work.' }
   planner:
     ready: { role: worker, prompt: 'Work on branch {{{branch}}} in {{{repoPath}}}' }
   worker:

packages/cli/src/__tests__/fixtures/e2e-suspend.workflow.yaml

@@ -18,7 +18,8 @@ roles:
           required: [$status]
 graph:
   $START:
-    _: { role: planner, prompt: 'Analyze the task' }
+    new: { role: planner, prompt: 'Analyze the task' }
+    resume: { role: planner, prompt: 'Review the previous run output and continue the work.' }
   planner:
     insufficient_info: { role: '$SUSPEND', prompt: 'Need more info: {{{reason}}}' }
     ready: { role: '$END', prompt: 'Done' }

packages/cli/src/__tests__/moderator-evaluate.test.ts

@@ -5,7 +5,12 @@ import { evaluate } from "../moderator/evaluate.js";
 
 const solveIssueGraph: WorkflowPayload["graph"] = {
   $START: {
-    _: { role: "planner", prompt: "Start planning from the issue in the task.", location: null },
+    new: { role: "planner", prompt: "Start planning from the issue in the task.", location: null },
+    resume: {
+      role: "planner",
+      prompt: "Review the previous run output and continue the work.",
+      location: null,
+    },
   },
   planner: {
     planned: { role: "developer", prompt: "Implement the plan: {{plan}}", location: null },
@@ -20,8 +25,8 @@ const solveIssueGraph: WorkflowPayload["graph"] = {
 };
 
 describe("evaluate", () => {
-  test("$START → first role (unit status _)", () => {
+  test("$START → first role (status new)", () => {
-    const result = evaluate(solveIssueGraph, "$START", { $status: "_" });
+    const result = evaluate(solveIssueGraph, "$START", { $status: "new" });
     expect(result).toEqual({
       ok: true,
       value: {
@@ -32,6 +37,18 @@ describe("evaluate", () => {
     });
   });
 
+  test("$START → first role (status resume)", () => {
+    const result = evaluate(solveIssueGraph, "$START", { $status: "resume" });
+    expect(result).toEqual({
+      ok: true,
+      value: {
+        role: "planner",
+        prompt: "Review the previous run output and continue the work.",
+        location: null,
+      },
+    });
+  });
+
   test("status-based routing (reviewer rejected → developer)", () => {
     const result = evaluate(solveIssueGraph, "reviewer", {
       $status: "rejected",
@@ -95,7 +112,7 @@ describe("evaluate", () => {
   });
 
   test("missing role in graph → error", () => {
-    const result = evaluate(solveIssueGraph, "unknown-role", { $status: "_" });
+    const result = evaluate(solveIssueGraph, "unknown-role", { $status: "new" });
     expect(result.ok).toBe(false);
     if (!result.ok) {
       expect(result.error.message).toBe('no transitions defined for role "unknown-role"');

packages/cli/src/__tests__/prompt.test.ts

@@ -9,31 +9,25 @@ import {
   cmdPromptAdapterDeveloping,
   cmdPromptBootstrap,
   cmdPromptList,
-  cmdPromptSetup,
   cmdPromptUsage,
-  cmdPromptUsageReference,
   cmdPromptWorkflowAuthoring,
 } from "../commands/prompt.js";
 
 describe("prompt commands", () => {
-  test("prompt list returns new prompt names", () => {
+  test("prompt list returns prompt names (no bootstrap)", () => {
     const result = cmdPromptList();
     expect(result).toBeInstanceOf(Array);
     expect(result).toContain("usage");
     expect(result).toContain("workflow-authoring");
     expect(result).toContain("adapter-developing");
-    expect(result).toContain("bootstrap");
+    expect(result).not.toContain("bootstrap");
-    expect(result).not.toContain("user");
-    expect(result).not.toContain("author");
-    expect(result).not.toContain("developer");
-    expect(result).not.toContain("adapter");
     for (const name of result) {
       expect(name).toMatch(/^\S+$/);
     }
   });
 
-  test("prompt usage-reference returns non-empty markdown string with frontmatter", () => {
+  test("prompt usage returns only the usage reference with frontmatter", () => {
-    const result = cmdPromptUsageReference();
+    const result = cmdPromptUsage();
     expect(typeof result).toBe("string");
     expect(result).toContain("uwf");
     expect(result).toContain("thread");
@@ -42,6 +36,9 @@ describe("prompt commands", () => {
     expect(result).toContain("---");
     expect(result).toContain("name:");
     expect(result).toContain("version:");
+    // Should NOT contain other references
+    expect(result).not.toContain("Workflow Authoring Reference");
+    expect(result).not.toContain("Adapter Developing Reference");
     expect(result.length).toBeGreaterThan(500);
   });
 
@@ -71,44 +68,29 @@ describe("prompt commands", () => {
     expect(result.length).toBeGreaterThan(500);
   });
 
-  test("prompt bootstrap returns non-empty skill with frontmatter", () => {
+  test("prompt bootstrap returns framework-agnostic setup instructions", () => {
     const result = cmdPromptBootstrap();
     expect(typeof result).toBe("string");
-    expect(result).toContain("uwf");
+    // Skills installation
-    expect(result).toContain("---");
-    expect(result.length).toBeGreaterThan(100);
-  });
-
-  test("prompt usage combines remaining references (no developer)", () => {
-    const result = cmdPromptUsage();
-    expect(typeof result).toBe("string");
-    expect(result).toContain("Usage Reference");
-    expect(result).toContain("Workflow Authoring Reference");
-    expect(result).toContain("Adapter Developing Reference");
-    expect(result).not.toContain("Developer Reference");
-    expect(result).toContain("---");
-    expect(result.length).toBeGreaterThan(2000);
-  });
-
-  test("prompt setup returns simplified setup instructions", () => {
-    const result = cmdPromptSetup();
-    expect(typeof result).toBe("string");
-    expect(result).toContain("uwf Skill Setup");
-    expect(result).toContain("uwf prompt bootstrap");
-    expect(result).toContain("SKILL.md");
-    expect(result).toContain("version");
-    expect(result).not.toMatch(/\bbun (install|run|test|changeset|version|release)\b/);
-  });
-
-  test("prompt setup references new subcommand names", () => {
-    const result = cmdPromptSetup();
     expect(result).toContain("uwf prompt usage");
     expect(result).toContain("uwf prompt workflow-authoring");
     expect(result).toContain("uwf prompt adapter-developing");
-    expect(result).not.toContain("uwf prompt user");
+    expect(result).toContain("uwf-usage");
-    expect(result).not.toContain("uwf prompt author");
+    expect(result).toContain("uwf-workflow-authoring");
-    expect(result).not.toContain("uwf prompt developer");
+    expect(result).toContain("uwf-adapter-developing");
-    expect(result).not.toMatch(/uwf prompt adapter\b(?!-developing)/);
+    // Fresh install scenario
+    expect(result).toContain("Fresh Install");
+    expect(result).toContain("uwf setup");
+    expect(result).toContain("--provider");
+    expect(result).toContain("--api-key");
+    expect(result).toContain("agent adapter");
+    // Upgrade scenario
+    expect(result).toContain("Upgrade");
+    expect(result).toContain("Migrate");
+    // Should NOT contain Hermes-specific paths
+    expect(result).not.toContain("~/.hermes/skills/");
+    expect(result).not.toContain("> ~/.hermes/");
+    expect(result.length).toBeGreaterThan(100);
   });
 
   test("prompt help subcommand is suppressed", { timeout: 30_000 }, () => {
@@ -119,11 +101,12 @@ describe("prompt commands", () => {
     });
     expect(output).not.toMatch(/help\s+\[command\]/i);
     expect(output).toContain("usage");
-    expect(output).toContain("setup");
+    expect(output).toContain("bootstrap");
     expect(output).toContain("workflow-authoring");
     expect(output).toContain("adapter-developing");
-    expect(output).toContain("bootstrap");
     expect(output).toContain("list");
-    expect(output).not.toContain("developer");
+    // Removed subcommands should not appear as command names
+    expect(output).not.toMatch(/^\s+setup\s/m);
+    expect(output).not.toContain("usage-reference");
   });
 });

packages/cli/src/__tests__/step-timing.test.ts

@@ -253,7 +253,10 @@ describe("thread read timing", () => {
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "go", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "go", location: null },
+          resume: { role: "worker", prompt: "resume", location: null },
+        },
         worker: { done: { role: "$END", prompt: "", location: null } },
       },
     });
@@ -319,7 +322,10 @@ describe("thread read timing", () => {
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "go", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "go", location: null },
+          resume: { role: "worker", prompt: "resume", location: null },
+        },
         worker: { done: { role: "$END", prompt: "", location: null } },
       },
     });

packages/cli/src/__tests__/thread-location.test.ts

@@ -57,10 +57,14 @@ roles:
         $status: { type: string, enum: ["ready"] }
 graph:
   $START:
-    _:
+    new:
       role: planner
       prompt: "Plan the work"
       location: null
+    resume:
+      role: planner
+      prompt: "Resume the work"
+      location: null
   planner:
     ready:
       role: $END
@@ -113,10 +117,14 @@ roles:
         $status: { type: string, enum: ["ready"] }
 graph:
   $START:
-    _:
+    new:
       role: planner
       prompt: "Plan"
       location: null
+    resume:
+      role: planner
+      prompt: "Resume"
+      location: null
   planner:
     ready:
       role: $END
@@ -156,10 +164,14 @@ roles:
         $status: { type: string, enum: ["ready"] }
 graph:
   $START:
-    _:
+    new:
       role: planner
       prompt: "Plan"
       location: null
+    resume:
+      role: planner
+      prompt: "Resume"
+      location: null
   planner:
     ready:
       role: $END

packages/cli/src/__tests__/thread-resume.test.ts

@@ -70,7 +70,10 @@ async function setupSuspendedThread(mode: MockAgentMode): Promise<{
       },
     },
     graph: {
-      $START: { _: { role: "worker", prompt: "Start work", location: null } },
+      $START: {
+        new: { role: "worker", prompt: "Start work", location: null },
+        resume: { role: "worker", prompt: "Resume the work", location: null },
+      },
       worker: {
         needs_input: {
           role: "$SUSPEND",
@@ -233,7 +236,10 @@ describe("uwf thread resume", () => {
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "Start", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "Start", location: null },
+          resume: { role: "worker", prompt: "Resume", location: null },
+        },
         worker: { done: { role: "$END", prompt: "Done", location: null } },
       },
     });
@@ -479,7 +485,10 @@ describe("uwf thread resume - completed threads", () => {
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "Start work", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "Start work", location: null },
+          resume: { role: "worker", prompt: "Resume the work", location: null },
+        },
         worker: { done: { role: "reviewer", prompt: "Review the work", location: null } },
         reviewer: { done: { role: "$END", prompt: "Done", location: null } },
       },
@@ -610,7 +619,7 @@ echo '${adapterJson}'
     expect(cliOutput.done).toBe(false);
 
     const capturedPrompt = await readFile(promptCapturePath, "utf8");
-    expect(capturedPrompt).toContain("Previous run completed");
+    expect(capturedPrompt).toContain("Resume the work");
     expect(capturedPrompt).toContain("Additional context");
 
     const storeModule = await import("../store.js");
@@ -640,7 +649,10 @@ echo '${adapterJson}'
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "Start", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "Start", location: null },
+          resume: { role: "worker", prompt: "Resume", location: null },
+        },
         worker: { done: { role: "$END", prompt: "Done", location: null } },
       },
     });
@@ -688,7 +700,10 @@ echo '${adapterJson}'
         },
       },
       graph: {
-        $START: { _: { role: "worker", prompt: "Start", location: null } },
+        $START: {
+          new: { role: "worker", prompt: "Start", location: null },
+          resume: { role: "worker", prompt: "Resume", location: null },
+        },
         worker: { done: { role: "$END", prompt: "Done", location: null } },
       },
     });

packages/cli/src/__tests__/thread-show-status.test.ts

@@ -34,10 +34,14 @@ roles:
         $status: { type: string, enum: ["ready"] }
 graph:
   $START:
-    _:
+    new:
       role: planner
       prompt: "Plan the work"
       location: null
+    resume:
+      role: planner
+      prompt: "Resume the work"
+      location: null
   planner:
     ready:
       role: $END
@@ -66,10 +70,14 @@ roles:
             question: { type: string }
 graph:
   $START:
-    _:
+    new:
       role: worker
       prompt: "Start work"
       location: null
+    resume:
+      role: worker
+      prompt: "Resume work"
+      location: null
   worker:
     needs_input:
       role: $SUSPEND

packages/cli/src/__tests__/thread-start-cwd-cli.test.ts

@@ -57,10 +57,14 @@ roles:
         $status: { type: string, enum: ["ready"] }
 graph:
   $START:
-    _:
+    new:
       role: planner
       prompt: "Plan the work"
       location: null
+    resume:
+      role: planner
+      prompt: "Resume the work"
+      location: null
   planner:
     ready:
       role: $END

packages/cli/src/__tests__/thread-suspend-step.test.ts

@@ -58,7 +58,10 @@ describe("suspend step CAS chain and threads.yaml metadata", () => {
           },
         },
         graph: {
-          $START: { _: { role: "worker", prompt: "Start work", location: null } },
+          $START: {
+            new: { role: "worker", prompt: "Start work", location: null },
+            resume: { role: "worker", prompt: "Resume work", location: null },
+          },
           worker: {
             needs_input: {
               role: "$SUSPEND",

packages/cli/src/__tests__/thread-suspended-display.test.ts

@@ -55,7 +55,10 @@ describe("suspended thread display", () => {
           },
         },
         graph: {
-          $START: { _: { role: "worker", prompt: "Start work", location: null } },
+          $START: {
+            new: { role: "worker", prompt: "Start work", location: null },
+            resume: { role: "worker", prompt: "Resume work", location: null },
+          },
           worker: {
             needs_input: {
               role: "$SUSPEND",
@@ -162,7 +165,10 @@ describe("suspended thread display", () => {
           },
         },
         graph: {
-          $START: { _: { role: "worker", prompt: "Start work", location: null } },
+          $START: {
+            new: { role: "worker", prompt: "Start work", location: null },
+            resume: { role: "worker", prompt: "Resume work", location: null },
+          },
           worker: {
             needs_input: {
               role: "$SUSPEND",
@@ -248,7 +254,10 @@ describe("suspended thread display", () => {
           },
         },
         graph: {
-          $START: { _: { role: "worker", prompt: "Start work", location: null } },
+          $START: {
+            new: { role: "worker", prompt: "Start work", location: null },
+            resume: { role: "worker", prompt: "Resume work", location: null },
+          },
         },
       });
 

packages/cli/src/__tests__/validate-semantic.test.ts

@@ -51,7 +51,10 @@ function makeWorkflow(overrides?: Partial<WorkflowPayload>): WorkflowPayload {
       },
     },
     graph: {
-      $START: { _: { role: "writer", prompt: "Begin writing", location: null } },
+      $START: {
+        new: { role: "writer", prompt: "Begin writing", location: null },
+        resume: { role: "writer", prompt: "Review previous output and continue", location: null },
+      },
       writer: { done: { role: "reviewer", prompt: "Review this: {{{plan}}}", location: null } },
       reviewer: {
         approved: { role: "$END", prompt: "Done: {{{summary}}}", location: null },
@@ -135,27 +138,38 @@ describe("Suite 2: Graph Structure", () => {
     expect(errors.some((e) => e.includes("$START must be defined in graph"))).toBe(true);
   });
 
-  test("2.2 $START has multiple status keys", () => {
+  test("2.2 $START missing resume edge", () => {
     const wf = makeWorkflow();
     wf.graph.$START = {
-      _: { role: "writer", prompt: "Begin", location: null },
+      new: { role: "writer", prompt: "Begin", location: null },
-      other: { role: "reviewer", prompt: "Also", location: null },
     };
     const errors = validateWorkflow(wf);
     expect(
-      errors.some((e) => e.includes('$START must have exactly one edge with status "_"')),
+      errors.some((e) => e.includes('$START must have edges with statuses "new" and "resume"')),
     ).toBe(true);
   });
 
-  test("2.3 $START edge uses non-_ status", () => {
+  test("2.3 $START missing new edge", () => {
     const wf = makeWorkflow();
-    wf.graph.$START = { ready: { role: "writer", prompt: "Begin", location: null } };
+    wf.graph.$START = {
+      resume: { role: "writer", prompt: "Resume", location: null },
+    };
     const errors = validateWorkflow(wf);
     expect(
-      errors.some((e) => e.includes('$START must have exactly one edge with status "_"')),
+      errors.some((e) => e.includes('$START must have edges with statuses "new" and "resume"')),
     ).toBe(true);
   });
 
+  test("2.3b $START with new and resume passes", () => {
+    const wf = makeWorkflow();
+    wf.graph.$START = {
+      new: { role: "writer", prompt: "Begin", location: null },
+      resume: { role: "writer", prompt: "Resume", location: null },
+    };
+    const errors = validateWorkflow(wf);
+    expect(errors.some((e) => e.includes("$START must have edges"))).toBe(false);
+  });
+
   test("2.4 $END has outgoing edges", () => {
     const wf = makeWorkflow();
     wf.graph.$END = { _: { role: "writer", prompt: "Loop", location: null } };
@@ -193,15 +207,18 @@ describe("Suite 2: Graph Structure", () => {
 });
 
 describe("Suite 3: Status-Edge Consistency", () => {
-  test("3.1 user role using _ graph key is rejected", () => {
+  test("3.1 user role using _ graph key is treated as an unknown status", () => {
+    // "_" is no longer special-cased — it's just a status key that does not
+    // match the role's $status enum, so it surfaces as extra/missing keys.
     const wf = makeWorkflow();
     wf.graph.writer = { _: { role: "reviewer", prompt: "Review", location: null } };
     const errors = validateWorkflow(wf);
-    expect(
+    expect(errors.some((e) => e.includes('role "writer" graph has extra status keys: _'))).toBe(
-      errors.some((e) =>
+      true,
-        e.includes('role "writer" must use explicit $status keys in graph, not "_"'),
+    );
-      ),
+    expect(errors.some((e) => e.includes('role "writer" graph is missing status keys: done'))).toBe(
-    ).toBe(true);
+      true,
+    );
   });
 
   test("3.2 user role graph key not matching $status enum", () => {
@@ -240,13 +257,16 @@ describe("Suite 3: Status-Edge Consistency", () => {
     ).toBe(true);
   });
 
-  test("3.5 multi-exit role with _ key", () => {
+  test("3.5 multi-exit role with _ key is treated as an unknown status", () => {
     const wf = makeWorkflow();
     wf.graph.reviewer = { _: { role: "$END", prompt: "Done", location: null } };
     const errors = validateWorkflow(wf);
+    expect(errors.some((e) => e.includes('role "reviewer" graph has extra status keys: _'))).toBe(
+      true,
+    );
     expect(
       errors.some((e) =>
-        e.includes('role "reviewer" must use explicit $status keys in graph, not "_"'),
+        e.includes('role "reviewer" graph is missing status keys: approved, rejected'),
       ),
     ).toBe(true);
   });

packages/cli/src/__tests__/workflow-resolution.test.ts

@@ -38,7 +38,10 @@ function makeMinimalPayload(name: string, description: string): WorkflowPayload
       },
     },
     graph: {
-      $START: { _: { role: "worker", prompt: "start working", location: null } },
+      $START: {
+        new: { role: "worker", prompt: "start working", location: null },
+        resume: { role: "worker", prompt: "resume working", location: null },
+      },
       worker: { done: { role: "$END", prompt: "done", location: null } },
     },
   };

packages/cli/src/cli.ts

@@ -8,12 +8,10 @@ import {
   cmdPromptAdapterDeveloping,
   cmdPromptBootstrap,
   cmdPromptList,
-  cmdPromptSetup,
   cmdPromptUsage,
-  cmdPromptUsageReference,
   cmdPromptWorkflowAuthoring,
 } from "./commands/prompt.js";
-import { cmdSetup, cmdSetupInteractive } from "./commands/setup.js";
+import { cmdSetup, cmdSetupInteractive, resolvePresetBaseUrl } from "./commands/setup.js";
 import { cmdStepFork, cmdStepList, cmdStepRead, cmdStepShow } from "./commands/step.js";
 import {
   cmdThreadCancel,
@@ -509,23 +507,16 @@ prompt.addHelpCommand(false);
 
 prompt
   .command("usage")
-  .description("Print the complete skill content (all references combined)")
+  .description("Print the usage reference (CLI guide + typical workflows)")
   .action(() => {
     console.log(cmdPromptUsage());
   });
 
 prompt
-  .command("setup")
+  .command("bootstrap")
-  .description("Print setup instructions for installing the uwf skill")
+  .description("Print setup instructions for installing uwf skills")
   .action(() => {
-    console.log(cmdPromptSetup());
+    console.log(cmdPromptBootstrap());
-  });
-
-prompt
-  .command("usage-reference")
-  .description("Print the usage reference (CLI guide + typical workflows)")
-  .action(() => {
-    console.log(cmdPromptUsageReference());
   });
 
 prompt
@@ -542,13 +533,6 @@ prompt
     console.log(cmdPromptAdapterDeveloping());
   });
 
-prompt
-  .command("bootstrap")
-  .description("Print the bootstrap skill YAML for Hermes agents")
-  .action(() => {
-    console.log(cmdPromptBootstrap());
-  });
-
 prompt
   .command("list")
   .description("List all available prompt names")
@@ -574,10 +558,14 @@ program
     }) => {
       const storageRoot = resolveStorageRoot();
       runAction(async () => {
-        if (opts.provider && opts.baseUrl && opts.apiKey && opts.model) {
+        // Resolve preset base-url when provider is known but --base-url is omitted
+        const resolvedBaseUrl =
+          opts.baseUrl ??
+          (opts.provider !== undefined ? resolvePresetBaseUrl(opts.provider) : null);
+        if (opts.provider && resolvedBaseUrl && opts.apiKey && opts.model) {
           const result = await cmdSetup({
             provider: opts.provider,
-            baseUrl: opts.baseUrl,
+            baseUrl: resolvedBaseUrl,
             apiKey: opts.apiKey,
             model: opts.model,
             agent: opts.agent ?? undefined,
@@ -588,7 +576,7 @@ program
           await cmdSetupInteractive(storageRoot);
         } else {
           throw new Error(
-            "Non-interactive setup requires all of: --provider, --base-url, --api-key, --model",
+            "Non-interactive setup requires: --provider, --api-key, --model (--base-url is optional for preset providers)",
           );
         }
       });

packages/cli/src/commands/prompt.ts

@@ -1,14 +1,35 @@
+import { readFileSync } from "node:fs";
+import { dirname, join } from "node:path";
+import { fileURLToPath } from "node:url";
 import {
   generateAdapterDevelopingReference,
-  generateBootstrapReference,
   generateUsageReference,
   generateWorkflowAuthoringReference,
 } from "@united-workforce/util";
 
+// CLI package version (for bootstrap prompt — uwf --version prints this)
+// Walk up from __dirname to find the nearest package.json (works from both src/ and dist/)
+function _findCliVersion(): string {
+  let dir = dirname(fileURLToPath(import.meta.url));
+  for (let i = 0; i < 5; i++) {
+    const candidate = join(dir, "package.json");
+    try {
+      const pkg = JSON.parse(readFileSync(candidate, "utf-8")) as { name?: string; version?: string };
+      if (pkg.name === "@united-workforce/cli") {
+        return pkg.version ?? "0.0.0";
+      }
+    } catch {
+      // not found, keep walking
+    }
+    dir = dirname(dir);
+  }
+  return "0.0.0";
+}
+const CLI_VERSION = _findCliVersion();
+
 export {
   generateAdapterDevelopingReference as cmdPromptAdapterDeveloping,
-  generateBootstrapReference as cmdPromptBootstrap,
+  generateUsageReference as cmdPromptUsage,
-  generateUsageReference as cmdPromptUsageReference,
   generateWorkflowAuthoringReference as cmdPromptWorkflowAuthoring,
 };
 
@@ -16,66 +37,205 @@ const PROMPT_ENTRIES: ReadonlyArray<{ name: string; generate: () => string }> =
   { name: "usage", generate: generateUsageReference },
   { name: "workflow-authoring", generate: generateWorkflowAuthoringReference },
   { name: "adapter-developing", generate: generateAdapterDevelopingReference },
-  { name: "bootstrap", generate: generateBootstrapReference },
 ];
 
 export function cmdPromptList(): ReadonlyArray<string> {
   return PROMPT_ENTRIES.map((e) => e.name);
 }
 
-export function cmdPromptUsage(): string {
+export function cmdPromptBootstrap(): string {
-  return PROMPT_ENTRIES.filter((e) => e.name !== "bootstrap")
+  return `# uwf Bootstrap
-    .map((e) => e.generate())
-    .join("\n\n---\n\n");
-}
 
-export function cmdPromptSetup(): string {
+Set up or upgrade uwf (United Workforce) — from zero to running your first workflow.
-  return `# uwf Skill Setup
 
-You are being asked to install or update the uwf (United Workforce) skill
+## Scenario A: Fresh Install
-so that you know how to use the \`uwf\` CLI for workflow orchestration.
 
-## Steps
+### Step 0 — Environment pre-flight check
 
-1. **Check if uwf CLI is installed:**
+Run ALL checks below BEFORE installing anything. If any check fails, fix it first — do NOT skip ahead.
-   \`\`\`bash
-   uwf --help
-   \`\`\`
-   If not installed: \`npm install -g @united-workforce/cli\`
-
-2. **Clean up old versions of the skill** (if any exist):
-   - Look for any existing uwf/workflow skill files in your skill directories
-     (e.g. \`~/.hermes/skills/uwf/SKILL.md\` or similar paths).
-   - If a skill exists, read its \`version\` from the YAML frontmatter.
-   - If the version matches the current CLI version (\`uwf --version\`), you're done — no update needed.
-   - If the version is older, **delete the old skill file** before proceeding.
-
-3. **Install the bootstrap skill:**
-   \`\`\`bash
-   uwf prompt bootstrap > ~/.hermes/skills/uwf/SKILL.md
-   \`\`\`
-   The bootstrap prompt already includes complete YAML frontmatter — no editing needed.
-
-4. **Verify** the skill is loadable by your agent framework.
-
-## Individual prompts
-
-Each prompt outputs a complete SKILL.md with frontmatter — pipe directly to a file:
 
 \`\`\`bash
-uwf prompt list                                              # list available prompt names
+# 1. Node.js
-uwf prompt usage > ~/.hermes/skills/uwf-usage/SKILL.md      # CLI usage guide
+node --version   # need v20+
-uwf prompt workflow-authoring > ~/.hermes/skills/uwf-workflow-authoring/SKILL.md
+# FIX: install via nvm (https://github.com/nvm-sh/nvm) or your OS package manager
-uwf prompt adapter-developing > ~/.hermes/skills/uwf-adapter-developing/SKILL.md
+
-uwf prompt bootstrap > ~/.hermes/skills/uwf/SKILL.md        # bootstrap skill
+# 2. Package manager — pnpm (preferred) or npm
+pnpm --version || npm --version
+# FIX for pnpm: corepack enable pnpm   (Node 20+)
+#   or: npm install -g pnpm
+# If pnpm is not an option, npm works too — adjust install commands below
+
+# 3. Global bin directory is in PATH
+# For pnpm:
+pnpm bin -g       # prints the global bin directory
+# For npm:
+npm prefix -g     # global prefix; bin is <prefix>/bin
+# The directory printed above MUST be in your PATH.
+# Test: echo \\$PATH | tr ':' '\\\\n'
+# FIX: add the missing directory to your shell's startup file
+#   (e.g. ~/.bashrc, ~/.zshrc, ~/.profile, or fish config):
+#   export PATH="<global-bin-dir>:$PATH"
+
+# 4. (uwf-hermes only) hermes CLI
+which hermes
+# FIX: if hermes is in a virtualenv, add it to PATH:
+#   export PATH="$HOME/.hermes/hermes-agent/.venv/bin:$PATH"
+#   or create a symlink: ln -s ~/.hermes/hermes-agent/.venv/bin/hermes ~/.local/bin/hermes
 \`\`\`
 
-## Notes
+**All checks must pass before continuing.** If you had to modify PATH, verify the change persists by opening a new shell or sourcing your shell config.
 
-- The skill content is bundled with the CLI and versioned with it — always use
+### Step 1 — Install CLI and agent adapter
-  \`uwf prompt usage\` to get the content matching your installed version.
+
-- Do NOT hand-edit the skill body. If the CLI is updated, re-run \`uwf prompt setup\`
+\`\`\`bash
-  and follow the steps again.
+pnpm add -g @united-workforce/cli    # or: npm install -g @united-workforce/cli
-- When upgrading, always delete the old skill first to avoid stale instructions.
+uwf --version   # should print ${CLI_VERSION}
+\`\`\`
+
+Install an agent adapter (at least one is required):
+
+| Adapter | Install | When to use |
+|---------|---------|-------------|
+| uwf-hermes | \`pnpm add -g @united-workforce/agent-hermes\` | When your agent framework is Hermes Agent |
+| uwf-claude-code | \`pnpm add -g @united-workforce/agent-claude-code\` | When using Claude Code CLI directly |
+| uwf-builtin | \`pnpm add -g @united-workforce/agent-builtin\` | Lightweight built-in agent (no external dependency) |
+
+**uwf-hermes** also requires the Hermes ACP plugin. After installing \`hermes-agent\`, run:
+\`\`\`bash
+pip install hermes-agent[acp]   # or: pip install -e .[acp] if installed from source
+\`\`\`
+
+Verify the adapter is installed: \`uwf-hermes --version\` (or whichever you chose).
+
+### Step 2 — Configure provider and model
+
+uwf needs an LLM provider to run agents. **Ask the user** for their provider, API key, and model, then run:
+
+\`\`\`bash
+uwf setup --provider <name> --base-url <url> --api-key <key> --model <model> [--agent <adapter>]
+\`\`\`
+
+Preset providers (base-url is auto-filled when using a preset name):
+openai, xai, openrouter, venice, dashscope, deepseek, siliconflow, volcengine, kimi, glm, stepfun, minimax, ollama
+
+Example:
+\`\`\`bash
+uwf setup --provider openrouter --api-key sk-or-... --model anthropic/claude-sonnet-4 --agent uwf-hermes
+\`\`\`
+
+If the user doesn't know what to choose, suggest \`openrouter\` with \`anthropic/claude-sonnet-4\` as a sensible default.
+
+Config is saved to \`~/.uwf/config.yaml\`. Verify with \`cat ~/.uwf/config.yaml\`.
+
+### Step 3 — Install skills
+
+Run each command and create a skill from its output:
+
+\`\`\`bash
+uwf prompt usage               # → save as skill "uwf-usage"
+uwf prompt workflow-authoring  # → save as skill "uwf-workflow-authoring"
+uwf prompt adapter-developing  # → save as skill "uwf-adapter-developing"
+\`\`\`
+
+Each command outputs a complete SKILL.md with YAML frontmatter. Use your agent framework's skill creation API to save them (e.g. \`skill_manage(action='create', name='uwf-usage', content=<output>)\`).
+
+Verify skills are installed by listing them (e.g. \`skills_list()\`) and confirming all three appear.
+
+### Step 4 — Verify end-to-end
+
+Create a minimal workflow file to test your setup:
+
+\`\`\`bash
+cat > /tmp/hello.yaml << 'YAML'
+name: hello
+description: Minimal smoke test
+roles:
+  greeter:
+    description: "Greet the user"
+    goal: "Respond with a friendly greeting"
+    capabilities: []
+    procedure: "Write a short greeting based on the prompt."
+    output: "A greeting message."
+    frontmatter:
+      type: object
+      properties:
+        $status: { enum: [done] }
+        message: { type: string }
+      required: [$status, message]
+graph:
+  $START:
+    new: { role: greeter, prompt: "Say hello to the user." }
+    resume: { role: greeter, prompt: "Greet the user again." }
+  greeter:
+    done: { role: "$END", prompt: "Done." }
+YAML
+\`\`\`
+
+Then run:
+
+\`\`\`bash
+uwf thread start /tmp/hello.yaml -p "Hello, world!"
+uwf thread exec <thread-id>
+uwf thread show <thread-id>
+\`\`\`
+
+If the thread reaches \`$END\` with status \`completed\`, the setup is working.
+
+## Scenario B: Upgrade from Previous Version
+
+### Step 1 — Update packages
+
+\`\`\`bash
+pnpm add -g @united-workforce/cli@latest    # or: npm install -g @united-workforce/cli@latest
+uwf --version   # should print ${CLI_VERSION}
+
+# Also update your adapter(s)
+pnpm add -g @united-workforce/agent-hermes@latest
+\`\`\`
+
+### Step 2 — Regenerate skills
+
+Skill content is bundled with the CLI — always regenerate after upgrading:
+
+\`\`\`bash
+uwf prompt usage               # → update skill "uwf-usage"
+uwf prompt workflow-authoring  # → update skill "uwf-workflow-authoring"
+uwf prompt adapter-developing  # → update skill "uwf-adapter-developing"
+\`\`\`
+
+### Step 3 — Migrate workflow YAML files (if needed)
+
+Check the changelog for breaking changes. Known migrations:
+
+- **v0.2.0**: \`$START._\` → \`$START.new\` + \`$START.resume\`. All workflow YAML files must be updated:
+  \`\`\`yaml
+  # Before (v0.1.x)
+  $START:
+    _: { role: planner, prompt: "..." }
+
+  # After (v0.2.0+)
+  $START:
+    new: { role: planner, prompt: "..." }
+    resume: { role: planner, prompt: "Review previous run and continue." }
+  \`\`\`
+
+Update all \`.workflow/\` and \`.workflows/\` YAML files in your projects. \`uwf workflow add\` will reject files with the old \`_\` syntax.
+
+### Step 4 — Verify
+
+\`\`\`bash
+uwf thread start <your-workflow> -p "upgrade test"
+uwf thread exec <thread-id>
+\`\`\`
+
+## Available prompts
+
+\`\`\`bash
+uwf prompt list                # list available prompt names
+uwf prompt usage               # CLI usage guide
+uwf prompt workflow-authoring  # workflow YAML design guide
+uwf prompt adapter-developing  # building agent adapters
+uwf prompt bootstrap           # this guide
+\`\`\`
 `;
 }

packages/cli/src/commands/setup.ts

@@ -1,3 +1,4 @@
+import { execFileSync } from "node:child_process";
 import { existsSync, mkdirSync, readdirSync, readFileSync, statSync, writeFileSync } from "node:fs";
 import { join } from "node:path";
 import { stdin as input, stdout as output } from "node:process";
@@ -72,6 +73,12 @@ const PRESET_PROVIDERS = [
   { name: "ollama", label: "Ollama (local)", baseUrl: "http://localhost:11434/v1" },
 ] as const;
 
+/** Look up the base URL for a preset provider name. Returns null if not a preset. */
+export function resolvePresetBaseUrl(providerName: string): string | null {
+  const preset = PRESET_PROVIDERS.find((p) => p.name === providerName);
+  return preset !== undefined ? preset.baseUrl : null;
+}
+
 type SetupArgs = {
   provider: string;
   baseUrl: string;
@@ -175,7 +182,6 @@ export async function _discoverAgents(): Promise<string[]> {
 
 async function _tryWhichDiscovery(): Promise<string[] | null> {
   try {
-    const { execFileSync } = await import("node:child_process");
     const text = execFileSync("which", ["-a", "uwf-hermes", "uwf-claude-code", "uwf-cursor"], {
       encoding: "utf-8",
       stdio: ["pipe", "pipe", "pipe"],
@@ -391,6 +397,37 @@ function mergeConfig(existing: Record<string, unknown>, args: SetupArgs): Record
   };
 }
 
+/**
+ * Check if the configured adapter binary (and its dependencies) are in PATH.
+ * Returns warnings array — empty means all good.
+ */
+export function _checkAdapterAvailability(agentName: string): string[] {
+  const warnings: string[] = [];
+  const binary = `uwf-${agentName}`;
+
+  try {
+    execFileSync("which", [binary], { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"] });
+  } catch {
+    warnings.push(
+      `${binary} not found in PATH. Install it: pnpm add -g @united-workforce/agent-${agentName}`,
+    );
+    return warnings; // skip dependency check if adapter itself is missing
+  }
+
+  // uwf-hermes depends on hermes CLI
+  if (agentName === "hermes") {
+    try {
+      execFileSync("which", ["hermes"], { encoding: "utf8", stdio: ["pipe", "pipe", "pipe"] });
+    } catch {
+      warnings.push(
+        'hermes CLI not found in PATH (required by uwf-hermes). Fix: export PATH="$HOME/.hermes/hermes-agent/.venv/bin:$PATH"',
+      );
+    }
+  }
+
+  return warnings;
+}
+
 /**
  * Non-interactive setup. All required args provided via CLI flags.
  */
@@ -405,15 +442,28 @@ export async function cmdSetup(args: SetupArgs): Promise<Record<string, unknown>
 
   writeFileSync(configPath, stringify(merged, { indent: 2 }), "utf8");
 
+  // Print config path to stderr (stdout is reserved for JSON output)
+  // biome-ignore lint/nursery/noConsole: CLI user-facing output
+  console.error(`Config saved to ${configPath} ✓`);
+
   // Validate model connectivity
   const validation = await validateModel(args.baseUrl, args.apiKey, args.model);
 
+  // Check adapter availability
+  const agentName = _agentNameFromBinary(args.agent ?? "hermes");
+  const adapterWarnings = _checkAdapterAvailability(agentName);
+  for (const w of adapterWarnings) {
+    // biome-ignore lint/nursery/noConsole: CLI user-facing output
+    console.error(`⚠ ${w}`);
+  }
+
   return {
     configPath,
     provider: args.provider,
     model: args.model,
     defaultAgent: merged.defaultAgent,
     validation,
+    adapterWarnings,
   };
 }
 

packages/cli/src/commands/thread.ts

@@ -911,7 +911,7 @@ function resolveEvaluateArgs(
   chain: ChainState,
 ): { lastRole: string; lastOutput: EvaluateLastOutput } {
   if (chain.headIsStart) {
-    return { lastRole: START_ROLE, lastOutput: { [STATUS_KEY]: "_" } };
+    return { lastRole: START_ROLE, lastOutput: { [STATUS_KEY]: "new" } };
   }
 
   const lastStep = chain.stepsNewestFirst[0];
@@ -1001,9 +1001,21 @@ function spawnAgent(
       stdio: ["ignore", "pipe", "pipe"],
       maxBuffer: 50 * 1024 * 1024, // 50 MB — stream-json output can be large
       cwd,
+      env: {
+        ...process.env,
+        NODE_OPTIONS: [process.env.NODE_OPTIONS, "--disable-warning=ExperimentalWarning"]
+          .filter(Boolean)
+          .join(" "),
+      },
     });
   } catch (e) {
     const err = e as NodeJS.ErrnoException & { stderr?: Buffer | string | null };
+    if (err.code === "ENOENT") {
+      failStep(
+        plog,
+        `"${agent.command}" not found in PATH. Install it or check your PATH config. Run: which ${agent.command}`,
+      );
+    }
     const stderr =
       err.stderr == null
         ? ""
@@ -1037,7 +1049,6 @@ function archiveThread(uwf: UwfStore, threadId: ThreadId, _workflow: CasRef, _he
   completeThread(uwf.varStore, threadId, "completed");
 }
 
-// biome-ignore lint/complexity/noExcessiveCognitiveComplexity: orchestration function with inherent branching
 export async function cmdThreadResume(
   storageRoot: string,
   threadId: ThreadId,
@@ -1101,7 +1112,7 @@ export async function cmdThreadResume(
 
   // status === "completed"
   const workflow = loadWorkflowPayload(uwf, workflowHash);
-  const startResult = evaluate(workflow.graph, START_ROLE, {});
+  const startResult = evaluate(workflow.graph, START_ROLE, { [STATUS_KEY]: "resume" });
   if (!startResult.ok) {
     fail(`failed to evaluate $START: ${startResult.error.message}`);
   }
@@ -1113,11 +1124,7 @@ export async function cmdThreadResume(
   }
 
   const startRole = startResult.value.role;
-  const completedPromptPrefix = "Previous run completed. Resuming with additional context.";
+  const completedResumePrompt = buildResumePrompt(startResult.value.prompt, supplement);
-  const completedResumePrompt =
-    supplement !== null && supplement !== ""
-      ? `${completedPromptPrefix}\n\n${supplement}`
-      : completedPromptPrefix;
 
   const updatedEntry = { ...entry, status: "idle" as const, completedAt: null };
   setThread(uwf.varStore, threadId, updatedEntry);
@@ -1247,6 +1254,12 @@ async function cmdThreadStepBackground(
   const child = spawn(scriptPath, args, {
     detached: true,
     stdio: "ignore",
+    env: {
+      ...process.env,
+      NODE_OPTIONS: [process.env.NODE_OPTIONS, "--disable-warning=ExperimentalWarning"]
+        .filter(Boolean)
+        .join(" "),
+    },
   });
 
   child.unref();

packages/cli/src/moderator/__tests__/evaluate.test.ts

@@ -6,11 +6,11 @@ describe("Edge prompt template variable resolution", () => {
   test("returns error when rendered prompt is empty string", () => {
     const graph = {
       $START: {
-        _: { role: "classifier", prompt: "{{{userPrompt}}}", location: null },
+        new: { role: "classifier", prompt: "{{{userPrompt}}}", location: null },
       },
     };
 
-    const result = evaluate(graph, "$START", {});
+    const result = evaluate(graph, "$START", { $status: "new" });
 
     expect(result.ok).toBe(false);
     if (!result.ok) {
@@ -22,11 +22,11 @@ describe("Edge prompt template variable resolution", () => {
   test("returns error when rendered prompt is whitespace-only", () => {
     const graph = {
       $START: {
-        _: { role: "classifier", prompt: "  {{{userPrompt}}}  ", location: null },
+        new: { role: "classifier", prompt: "  {{{userPrompt}}}  ", location: null },
       },
     };
 
-    const result = evaluate(graph, "$START", {});
+    const result = evaluate(graph, "$START", { $status: "new" });
 
     expect(result.ok).toBe(false);
     if (!result.ok) {
@@ -38,11 +38,11 @@ describe("Edge prompt template variable resolution", () => {
   test("succeeds when all template variables resolve to non-empty values", () => {
     const graph = {
       $START: {
-        _: { role: "classifier", prompt: "{{{userPrompt}}}", location: null },
+        new: { role: "classifier", prompt: "{{{userPrompt}}}", location: null },
       },
     };
 
-    const result = evaluate(graph, "$START", { userPrompt: "Fix the bug" });
+    const result = evaluate(graph, "$START", { $status: "new", userPrompt: "Fix the bug" });
 
     expect(result.ok).toBe(true);
     if (result.ok) {
@@ -53,11 +53,11 @@ describe("Edge prompt template variable resolution", () => {
   test("succeeds with static (no-variable) prompt", () => {
     const graph = {
       $START: {
-        _: { role: "classifier", prompt: "Classify this input", location: null },
+        new: { role: "classifier", prompt: "Classify this input", location: null },
       },
     };
 
-    const result = evaluate(graph, "$START", {});
+    const result = evaluate(graph, "$START", { $status: "new" });
 
     expect(result.ok).toBe(true);
     if (result.ok) {
@@ -68,11 +68,11 @@ describe("Edge prompt template variable resolution", () => {
   test("succeeds when prompt has mix of static text and unresolved variables", () => {
     const graph = {
       $START: {
-        _: { role: "classifier", prompt: "Please handle: {{{userPrompt}}}", location: null },
+        new: { role: "classifier", prompt: "Please handle: {{{userPrompt}}}", location: null },
       },
     };
 
-    const result = evaluate(graph, "$START", {});
+    const result = evaluate(graph, "$START", { $status: "new" });
 
     expect(result.ok).toBe(true);
     if (result.ok) {
@@ -83,11 +83,11 @@ describe("Edge prompt template variable resolution", () => {
   test("returns error when ALL variables missing and no static text remains", () => {
     const graph = {
       $START: {
-        _: { role: "classifier", prompt: "{{{a}}}{{{b}}}", location: null },
+        new: { role: "classifier", prompt: "{{{a}}}{{{b}}}", location: null },
       },
     };
 
-    const result = evaluate(graph, "$START", {});
+    const result = evaluate(graph, "$START", { $status: "new" });
 
     expect(result.ok).toBe(false);
   });

packages/cli/src/moderator/evaluate.ts

@@ -6,10 +6,7 @@ import type { EvaluateResult, Result } from "./types.js";
 // Disable HTML escaping — prompts are plain text, not HTML.
 mustache.escape = (text: string) => text;
 
-const START_ROLE = "$START";
 const SUSPEND_ROLE = "$SUSPEND";
-// $START is a special entry node with no agent output — it always uses this key.
-const START_STATUS = "_";
 
 type LastOutput = Record<string, unknown>;
 
@@ -21,9 +18,7 @@ export function evaluate(
   lastOutput: LastOutput,
 ): Result<EvaluateResult, Error> {
   let status: string;
-  if (lastRole === START_ROLE) {
+  if (typeof lastOutput[STATUS_KEY] === "string") {
-    status = START_STATUS;
-  } else if (typeof lastOutput[STATUS_KEY] === "string") {
     status = lastOutput[STATUS_KEY] as string;
   } else {
     return {

packages/cli/src/validate-semantic.ts

@@ -97,9 +97,9 @@ function checkGraphStructure(payload: WorkflowPayload, errors: string[]): void {
   if (!graphNodes.has("$START")) {
     errors.push("$START must be defined in graph");
   } else {
-    const startKeys = Object.keys(payload.graph.$START);
+    const startKeys = new Set(Object.keys(payload.graph.$START));
-    if (startKeys.length !== 1 || startKeys[0] !== "_") {
+    if (!startKeys.has("new") || !startKeys.has("resume")) {
-      errors.push('$START must have exactly one edge with status "_"');
+      errors.push('$START must have edges with statuses "new" and "resume"');
     }
   }
 
@@ -190,22 +190,13 @@ function checkOneOfDiscriminant(
   }
 }
 
-/** Check status-edge consistency for a user role. "_" is reserved for $START and rejected here. */
+/** Check status-edge consistency for a user role. */
 function checkStatusEdges(
   roleName: string,
   graphKeys: Set<string>,
   statusSet: Set<string>,
   errors: string[],
 ): void {
-  if (graphKeys.has("_")) {
-    errors.push(`role "${roleName}" must use explicit $status keys in graph, not "_"`);
-    return;
-  }
-  if (statusSet.has("_")) {
-    errors.push(`role "${roleName}" $status enum must use explicit values, not "_"`);
-    return;
-  }
-
   const extraKeys = [...graphKeys].filter((k) => !statusSet.has(k));
   const missingKeys = [...statusSet].filter((k) => !graphKeys.has(k));
   if (extraKeys.length > 0) {

packages/cli/src/validate.ts

@@ -57,13 +57,13 @@ function isGraph(value: unknown): boolean {
   if (!isRecord(value)) {
     return false;
   }
-  return Object.entries(value).every(([node, statusMap]) => {
+  return Object.values(value).every((statusMap) => {
     if (!isRecord(statusMap)) {
       return false;
     }
     return Object.entries(statusMap).every(([status, target]) => {
-      // "_" is only valid as a status key for the $START entry node.
+      // "_" is no longer a valid status key anywhere — $START uses "new"/"resume".
-      if (status === "_" && node !== "$START") {
+      if (status === "_") {
         return false;
       }
       return isTarget(target);
@@ -99,7 +99,7 @@ export function checkWorkflowFilenameConsistency(
 ): string | null {
   const expected = workflowNameFromPath(filePath);
   if (payload.name !== expected) {
-    return `workflow name mismatch: file "${basename(filePath)}" implies name "${expected}" but YAML declares name "${payload.name}"`;
+    return `workflow name mismatch: file "${basename(filePath)}" implies name "${expected}" but YAML declares name "${payload.name}". Either rename the file to "${payload.name}.yaml" or change the YAML \`name\` field to "${expected}"`;
   }
   return null;
 }

packages/eval/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/eval",
-  "version": "0.1.2",
+  "version": "0.1.3",
   "private": false,
   "files": [
     "src",

packages/eval/src/cli.ts

@@ -7,12 +7,15 @@ import {
   registerRunCommand,
 } from "./commands/index.js";
 
+// eslint-disable-next-line -- dynamic import for version
+const pkg = await import("../package.json", { with: { type: "json" } });
+
 const program = new Command();
 
 program
   .name("uwf-eval")
   .description("Evaluate uwf workflow quality with real agents")
-  .version("0.1.0");
+  .version(pkg.default.version, "-V, --version");
 
 registerRunCommand(program);
 registerReportCommand(program);

packages/util/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@united-workforce/util",
-  "version": "0.1.0",
+  "version": "0.1.2",
   "files": [
     "src",
     "dist",

packages/util/src/bootstrap-reference.ts

@@ -1,40 +0,0 @@
-export function generateBootstrapReference(): string {
-  return `---
-name: uwf
-description: "United Workforce (uwf) — YAML 状态机工作流引擎。任务涉及 workflow 时加载此 skill。"
-tags: [workflow, uwf]
-triggers:
-  - uwf
-  - workflow
-  - 工作流
----
-
-# uwf (United Workforce)
-
-YAML 状态机工作流引擎。当用户提到「workflow」「工作流」时，指的是 **uwf workflow**（YAML 定义的状态机），不是 Hermes skill。用 \`uwf\` CLI 操作，不要混淆。
-
-## 首次使用
-
-运行以下命令获取完整用法：
-
-\`\`\`bash
-uwf prompt usage                 # 完整用法文档（所有引用合并）
-uwf prompt workflow-authoring    # workflow 编写指南（role 定义、graph 路由、schema）
-uwf prompt adapter-developing    # adapter 开发指南（构建新的 agent adapter）
-\`\`\`
-
-## 快速参考
-
-\`\`\`bash
-uwf workflow list                          # 查看已注册 workflow
-uwf workflow add <file.yaml>               # 注册 workflow
-uwf thread start <workflow> -p "prompt"    # 创建 thread
-uwf thread exec <thread-id> -c 10          # 执行最多 10 步
-uwf thread list                            # 查看所有 thread
-\`\`\`
-
-## 示例 workflow
-
-参考项目 \`examples/\` 目录下的 YAML 文件（analyze-topic、debate、solve-issue）。
-`;
-}

packages/util/src/index.ts

@@ -2,7 +2,6 @@ export { generateActorReference } from "./actor-reference.js";
 export { generateAdapterDevelopingReference } from "./adapter-developing-reference.js";
 export { generateArchitectureReference } from "./architecture-reference.js";
 export { encodeUint64AsCrockford } from "./base32.js";
-export { generateBootstrapReference } from "./bootstrap-reference.js";
 export { generateCliReference } from "./cli-reference.js";
 export { env } from "./env.js";
 export type {
@@ -16,7 +15,7 @@ export {
   validateFrontmatter,
 } from "./frontmatter-markdown/index.js";
 export { createLogger } from "./logger.js";
-export { generateModeratorReference } from "./moderator-reference.js";
+
 export type {
   CreateProcessLoggerOptions,
   ProcessLogFn,
@@ -36,4 +35,3 @@ export { extractUlidTimestamp, generateUlid } from "./ulid.js";
 export { generateUsageReference } from "./usage-reference.js";
 export { VERSION } from "./version.js";
 export { generateWorkflowAuthoringReference } from "./workflow-authoring-reference.js";
-export { generateYamlReference } from "./yaml-reference.js";

packages/util/src/moderator-reference.ts

@@ -1,56 +0,0 @@
-export function generateModeratorReference(): string {
-  return `# Moderator Reference
-
-## Overview
-
-The moderator is the workflow engine's routing component. It evaluates the directed graph defined in the workflow YAML to determine the next role (or \`$END\`) after each step — with zero LLM cost.
-
-## Status-Based Routing
-
-The moderator uses **status-based routing**: it inspects the previous step's extracted output (specifically the \`$status\` field) and looks up the corresponding edge in the graph.
-
-### Graph Structure
-
-The graph is a nested map: \`Record<Role | "$START", Record<Status, Target>>\`. Each role maps its possible \`$status\` values to a target with a \`role\` and \`prompt\`:
-
-\`\`\`yaml
-graph:
-  $START:
-    _: { role: planner, prompt: "Analyze the issue." }
-  planner:
-    ready: { role: developer, prompt: "Implement the plan (CAS hash: {{{plan}}})." }
-    insufficient_info: { role: $END, prompt: "Not enough info." }
-  developer:
-    done: { role: reviewer, prompt: "Review branch {{{branch}}} at {{{worktree}}}." }
-    failed: { role: $END, prompt: "Developer failed: {{{reason}}}." }
-  reviewer:
-    approved: { role: tester, prompt: "Run tests on {{{branch}}} at {{{worktree}}}." }
-    rejected: { role: developer, prompt: "Fix issues: {{{comments}}}." }
-\`\`\`
-
-### Routing Algorithm
-
-1. Look up \`graph[lastRole]\` to get the status map for the current role
-2. Look up \`statusMap[lastOutput.$status]\` to get the target
-3. If target role is \`$END\`, mark thread as completed
-4. Otherwise, render the edge prompt (Mustache templates with \`{{{field}}}\` from output) and spawn the next agent
-
-### Edge Prompts and Mustache Templates
-
-Edge prompts use triple-brace Mustache syntax (\`{{{field}}}\`) to interpolate values from the previous step's output into the next agent's task prompt. This passes structured data (branch names, file paths, CAS hashes) between roles without manual wiring.
-
-## Special Nodes
-
-- \`$START\` — entry point; uses status key \`_\` (unconditional) since there is no previous output
-- \`$END\` — terminal node; thread completes when reached and is moved to history
-
-## Integration with Steps
-
-Each \`uwf thread exec\` cycle:
-1. Moderator reads the thread's head step output
-2. Looks up \`graph[lastRole][output.$status]\` to pick the next role
-3. If next is \`$END\`, marks thread as completed
-4. Otherwise, renders the edge prompt and spawns the agent for the selected role
-5. Extract pipeline parses agent output → new step node → append to CAS chain
-`;
-}

packages/util/src/version.ts

@@ -1,2 +1,9 @@
-// This version is kept in sync with package.json during releases.
+import { readFileSync } from "node:fs";
-export const VERSION = "0.1.0";
+import { dirname, join } from "node:path";
+import { fileURLToPath } from "node:url";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const pkg = JSON.parse(readFileSync(join(__dirname, "..", "package.json"), "utf-8")) as {
+  version: string;
+};
+export const VERSION = pkg.version;

packages/util/src/workflow-authoring-reference.ts

@@ -40,7 +40,8 @@ roles:                         # named actors
 
 graph:                         # status-based routing
   $START:
-    _: { role: planner, prompt: "Analyze the issue." }
+    new: { role: planner, prompt: "Analyze the issue." }
+    resume: { role: planner, prompt: "Review the previous run output and continue." }
   planner:
     ready: { role: developer, prompt: "Implement {{{plan}}}." }
     failed: { role: $END, prompt: "Failed: {{{error}}}" }
@@ -91,16 +92,21 @@ Add any fields you need for data passing between roles. These are available in e
 
 ### Flat Schema (Single Status)
 
-When a role has only one outcome:
+When a role has only one outcome, use \`enum\` with a single value:
 
 \`\`\`yaml
 frontmatter:
+  type: object
   properties:
-    $status: { const: "done" }
+    $status:
+      type: string
+      enum: [done]
     summary: { type: string }
   required: [$status, summary]
 \`\`\`
 
+Note: \`$status: { const: "done" }\` is **not** valid in flat schemas — the validator requires \`enum\` or \`oneOf\` with \`const\`. Use \`const\` only inside \`oneOf\` variants.
+
 ## Graph Routing
 
 The graph maps each role's \`$status\` values to the next role:
@@ -113,7 +119,7 @@ graph[role][$status] → { role: nextRole, prompt: edgePrompt }
 
 | Node | Purpose |
 |------|---------|
-| \`$START\` | Entry point — status key is always \`_\` (unconditional) |
+| \`$START\` | Entry point — status keys \`new\` (first start) and \`resume\` (resuming a completed thread) |
 | \`$END\` | Terminal — thread completes and is archived |
 
 ### Edge Prompts
@@ -178,7 +184,7 @@ ocas get <output-hash>
 1. Every \`$status\` value in a role's frontmatter has a matching edge in the graph
 2. Every field referenced in edge prompts (\`{{{field}}}\`) exists in the source role's schema
 3. Every role referenced in the graph exists in \`roles\`
-4. \`$START\` has exactly one edge with key \`_\`
+4. \`$START\` has edges with keys \`new\` and \`resume\`
 5. At least one path leads to \`$END\`
 6. No orphan roles (defined but never routed to)
 

packages/util/src/yaml-reference.ts

@@ -1,82 +0,0 @@
-export function generateYamlReference(): string {
-  return `# Workflow YAML Schema Reference
-
-## Top-Level Structure
-
-A workflow YAML file defines the complete workflow specification:
-
-\`\`\`yaml
-name: solve-issue          # verb-first kebab-case identifier
-description: "..."         # human-readable description
-
-roles:                     # named actors in the workflow
-  planner:
-    description: "Analyzes issue and outputs a plan"
-    goal: "You are a planning agent."
-    capabilities:
-      - issue-analysis
-      - planning
-    procedure: |
-      1. Read the issue
-      2. Produce a test spec
-    output: "Output the plan summary. Set $status to ready or insufficient_info."
-    frontmatter:           # JSON Schema for structured output (drives routing)
-      oneOf:
-        - properties:
-            $status: { const: ready }
-            plan: { type: string }
-          required: [$status, plan]
-        - properties:
-            $status: { const: insufficient_info }
-          required: [$status]
-
-graph:                     # status-based routing (nested map)
-  $START:
-    _: { role: planner, prompt: "Analyze the issue." }
-  planner:
-    ready: { role: developer, prompt: "Implement plan {{{plan}}}." }
-    insufficient_info: { role: $END, prompt: "Not enough info." }
-\`\`\`
-
-## roles
-
-Each role defines an actor in the workflow:
-
-| Field | Type | Description |
-|-------|------|-------------|
-| \`description\` | string | Short description of the role's purpose |
-| \`goal\` | string | System-level goal statement for the agent |
-| \`capabilities\` | string[] | Tags describing what the role can do |
-| \`procedure\` | string | Step-by-step instructions for the agent |
-| \`output\` | string | Description of expected output format |
-| \`frontmatter\` | JSON Schema | Defines the structured output the agent must produce |
-
-### frontmatter
-
-The \`frontmatter\` field is a standard JSON Schema object. The extract pipeline validates agent output against it. Key conventions:
-- \`$status\` field drives routing decisions in the graph
-- Use \`const\` or \`enum\` to constrain status values
-- Use \`oneOf\` to define multiple valid output shapes (one per status)
-- All \`required\` fields must appear in the agent's frontmatter output
-
-## graph
-
-The graph is a nested map defining status-based routing:
-
-\`\`\`
-Record<Role | "$START", Record<Status, { role: string, prompt: string }>>
-\`\`\`
-
-| Level | Key | Value |
-|-------|-----|-------|
-| Outer | Role name or \`$START\` | Status map for that role |
-| Inner | \`$status\` value (or \`_\` for unconditional) | Target: \`{ role, prompt }\` |
-
-### Special Nodes
-- \`$START\` — entry point; uses status key \`_\` (unconditional, no previous output)
-- \`$END\` — terminal node; thread completes when reached
-
-### Edge Prompts
-Prompts use triple-brace Mustache templates (\`{{{field}}}\`) to interpolate values from the previous step's output. Example: \`"Implement plan {{{plan}}} in repo {{{repoPath}}}."\`
-`;
-}

workflows/normalize-bun-monorepo.yaml

@@ -21,9 +21,12 @@ graph:
       role: package-metadata
       prompt: Biome setup failed ({{{reason}}}), but continue. Standardize package metadata for repo at {{{repoPath}}}.
   $START:
-    _:
+    new:
       role: workspace
       prompt: Set up bun workspace structure for repo at {{{repoPath}}}.
+    resume:
+      role: workspace
+      prompt: Review the previous run output and continue setting up the bun workspace structure for repo at {{{repoPath}}}.
   release:
     done:
       role: testing

workflows/solve-issue.yaml

@@ -283,9 +283,12 @@ roles:
         - error
 graph:
   $START:
-    _:
+    new:
       role: planner
       prompt: Analyze the issue and produce an implementation plan.
+    resume:
+      role: planner
+      prompt: Review the previous run output and continue the work.
   planner:
     insufficient_info:
       role: $SUSPEND