xiaoju
d310d43ab8
Store the fully assembled prompt sent to each agent in CAS as a text
node, referenced from StepNodePayload.assembledPrompt. This enables
exact reproduction of what the agent received for debugging hallucinations.
Changes:
- workflow-protocol: StepRecord + STEP_NODE_SCHEMA add assembledPrompt field
- workflow-util-agent: AgentRunResult includes assembledPrompt, run.ts stores it
- workflow-util-agent: schemas register TEXT_SCHEMA for prompt storage
- workflow-agent-claude-code: return assembled prompt from buildClaudeCodePrompt
- workflow-agent-hermes: return assembled prompt from buildHermesPrompt
- workflow-agent-builtin: return empty prompt (no prompt assembly)
- cli-workflow: step read --prompt renders the stored prompt
- All test fixtures updated for new field
Legacy steps without assembledPrompt show 'Prompt not recorded' message.
小橘 🍊
74 lines
3.1 KiB
TypeScript
74 lines
3.1 KiB
TypeScript
import { createMemoryStore, putSchema } from "@uncaged/json-cas";
|
|
import { describe, expect, test } from "vitest";
|
|
|
|
import { tryFrontmatterFastPath } from "../src/frontmatter.js";
|
|
|
|
// ── Helpers ───────────────────────────────────────────────────────────────────
|
|
|
|
const PLANNER_SCHEMA = {
|
|
type: "object",
|
|
properties: {
|
|
$status: { type: "string", enum: ["ready", "failed"] },
|
|
plan: { type: "string" },
|
|
},
|
|
required: ["$status"],
|
|
additionalProperties: false,
|
|
};
|
|
|
|
describe("adapter-stdout: A4 retry loop survives JSON output", () => {
|
|
test("A4. first extraction fails, second succeeds — final result has correct data", async () => {
|
|
const store = createMemoryStore();
|
|
const schemaHash = await putSchema(store, PLANNER_SCHEMA);
|
|
|
|
// Simulate the retry loop from createAgent (run.ts lines 163-173):
|
|
// First attempt: agent outputs garbage (no frontmatter)
|
|
const badOutput = "Here is my response without frontmatter.\nJust plain text.";
|
|
const firstAttempt = await tryFrontmatterFastPath(badOutput, schemaHash, store);
|
|
expect(firstAttempt).toBeNull();
|
|
|
|
// Second attempt (after correction message): agent outputs valid frontmatter
|
|
const goodOutput = `---\n$status: ready\nplan: corrected-hash\n---\nCorrected body with valid frontmatter.`;
|
|
const secondAttempt = await tryFrontmatterFastPath(goodOutput, schemaHash, store);
|
|
|
|
expect(secondAttempt).not.toBeNull();
|
|
expect(secondAttempt!.outputHash).toMatch(/^[0-9A-Z]{13}$/);
|
|
expect(secondAttempt!.frontmatter).toEqual({ $status: "ready", plan: "corrected-hash" });
|
|
expect(secondAttempt!.body).toBe("Corrected body with valid frontmatter.");
|
|
|
|
// Verify the final AdapterOutput shape would be correct
|
|
const adapterOutput = {
|
|
stepHash: "MOCK_STEP_HASH",
|
|
detailHash: "MOCK_DETAIL_HA",
|
|
role: "planner",
|
|
frontmatter: secondAttempt!.frontmatter,
|
|
body: secondAttempt!.body,
|
|
startedAtMs: 1000,
|
|
completedAtMs: 2000,
|
|
assembledPrompt: null,
|
|
};
|
|
|
|
const json = JSON.stringify(adapterOutput);
|
|
const parsed = JSON.parse(json);
|
|
expect(parsed.frontmatter).toEqual({ $status: "ready", plan: "corrected-hash" });
|
|
expect(parsed.body).toBe("Corrected body with valid frontmatter.");
|
|
expect(parsed.completedAtMs).toBeGreaterThanOrEqual(parsed.startedAtMs);
|
|
});
|
|
|
|
test("A4. all retries fail — extraction returns null on every attempt", async () => {
|
|
const store = createMemoryStore();
|
|
const schemaHash = await putSchema(store, PLANNER_SCHEMA);
|
|
|
|
const MAX_RETRIES = 2;
|
|
const badOutput = "No frontmatter here";
|
|
|
|
// Simulate MAX_FRONTMATTER_RETRIES iterations all failing
|
|
let extracted = await tryFrontmatterFastPath(badOutput, schemaHash, store);
|
|
for (let retry = 0; retry < MAX_RETRIES && extracted === null; retry++) {
|
|
// Each retry also gets bad output
|
|
extracted = await tryFrontmatterFastPath(badOutput, schemaHash, store);
|
|
}
|
|
|
|
expect(extracted).toBeNull();
|
|
});
|
|
});
|