Files
AgentCompany/tests/integration/autonomous-chain/ceo-phased-prompts.spec.ts
T
ZacharyZhang-NY ff5e22092a refactor(session): agent-scope rotation + accumulation; auto-register file artifacts
Sessions are already agent-scoped post-v34, but the code kept routing
rotation / accumulation / handoff through the task indirection. Heartbeat
runs (no task) silently skipped all three, letting taskless transcripts
grow past the model's context window — the 7M-token explosion.

- DB adds getSessionById(sessionId) — direct lookup bypassing
  tasks.active_session_id
- session-management-service: generateHandoff(session) is now agent-scoped,
  walking recent runs across tasks + heartbeats; getHandoffForAgent(agentId)
  is the canonical lookup, getHandoffContext(taskId) is a thin shim for
  callers that still hold a task id
- turn-context.ts::resolveSession drops the `if (!task)` early return:
  rotation evaluation runs on the agent session unconditionally
- run-finished-helpers.ts::accumulateAndRotateSession drops the `!task`
  gate: tokens accumulate and rotation fires for every run, keyed on the
  agent session, not the task
- Rotation evaluator contract: evaluatePreRunSessionRotation(agentId,
  connectorId) → { rotated, handoff, sessionId }; old (taskId, connectorId)
  signature removed

IPC contract tightening: ipc-support::emitEvent becomes a best-effort
sink. "Render frame was disposed" during renderer reload is a lifecycle
signal, not a logic error; silently drop it instead of letting it abort
downstream event emissions.

Artifact projection: pi's write/edit tool results now upsert a
`work_products` row via a new agent-runtime/artifact-projector.ts. No
more agents writing 54 files into the workspace that the UI never sees.
Migration v40 relaxes work_products.task_id to nullable so heartbeat
artifacts can register too.

File size: split agent-event-projection.ts out of runtime-bridge.ts to
keep the bridge under the 400-line cap after adding the projector hook.

Tests: 604 unit + integration green; generateHandoff callsites updated
for the single-argument signature; artifact-projector has pure-function
coverage via the projector module's own shape.
2026-04-18 21:37:56 -04:00

172 lines
6.0 KiB
TypeScript

/**
* Step 2: CEO phased prompts + pending approvals + credential escalation.
* Physically extracted from autonomous-chain.spec.ts without behavior changes.
*/
import { describe, expect, it } from "vitest";
import { buildSystemPrompt } from "../../../src/main/prompt-builder";
import type { ApprovalRecord } from "../../../src/shared/types";
import { installSharedSetup } from "./shared-setup";
const shared = installSharedSetup();
const { createAgent, buildCtx } = shared;
describe("Step 2: CEO phased prompts", () => {
it("Phase 0: shows team building instructions when no reports at all", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const ctx = buildCtx({
agentId: ceoId,
directReports: [], // truly no reports
chainOfCommand: [],
pendingApprovals: [],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Team Building Phase");
// The rewritten prompt uses "Design a minimal org" instead of the
// old "Design the Organization Structure" header. Check for both
// the stable Step-2 marker and the new ordering guarantee that
// blocks the self-loop the earlier prompt enabled.
expect(prompt).toContain("Design a minimal org");
expect(prompt).toContain("HIRES FIRST. STRATEGY APPROVAL LAST.");
expect(prompt).toContain("Submit org strategy");
expect(prompt).not.toContain("Awaiting Org Approval");
expect(prompt).not.toContain("Goal Decomposition");
});
it("Phase 0.5: shows awaiting hire approvals when pending_approval hires exist but no org approval", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const hireId = createAgent({ name: "CTO", role: "CTO", reportsTo: ceoId, status: "pending_approval" as "idle" });
const hire = shared.db.getAgent(hireId);
const ctx = buildCtx({
agentId: ceoId,
directReports: [hire],
chainOfCommand: [],
pendingApprovals: [],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Awaiting Hire Approvals");
expect(prompt).not.toContain("Team Building Phase");
expect(prompt).not.toContain("Goal Decomposition");
});
it("Phase 1: shows awaiting org approval when pending approve_ceo_strategy exists", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const ctx = buildCtx({
agentId: ceoId,
directReports: [],
chainOfCommand: [],
pendingApprovals: [
{
id: "approval-1",
companyId: shared.companyId,
type: "approve_ceo_strategy",
state: "pending",
payloadSummary: "Org structure: CEO + CTO + CMO",
impactSummary: "Team of 3",
requestedByAgentId: ceoId,
relatedTaskId: null,
decisionNote: "",
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
} as ApprovalRecord,
],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Awaiting Org Approval");
expect(prompt).not.toContain("Team Building Phase");
expect(prompt).not.toContain("Goal Decomposition");
});
it("Phase 2: shows normal executive instructions when active reports and no pending org approval", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const ctoId = createAgent({ name: "CTO", role: "CTO", reportsTo: ceoId, status: "idle" });
const cto = shared.db.getAgent(ctoId);
const ctx = buildCtx({
agentId: ceoId,
directReports: [cto],
chainOfCommand: [],
pendingApprovals: [],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Executive / CEO");
expect(prompt).toContain("Goal Decomposition");
expect(prompt).toContain("Delegation");
expect(prompt).toContain("Unblock & Coordinate");
expect(prompt).not.toContain("Team Building Phase");
expect(prompt).not.toContain("Awaiting Org Approval");
});
it("includes pending approvals section in prompt", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const ctoId = createAgent({ name: "CTO", role: "CTO", reportsTo: ceoId, status: "idle" });
const cto = shared.db.getAgent(ctoId);
const ctx = buildCtx({
agentId: ceoId,
directReports: [cto],
chainOfCommand: [],
pendingApprovals: [
{
id: "approval-x",
companyId: shared.companyId,
type: "secret_access",
state: "pending",
payloadSummary: "Need Twitter API keys",
impactSummary: "",
requestedByAgentId: ceoId,
relatedTaskId: null,
decisionNote: "",
createdAt: new Date().toISOString(),
updatedAt: new Date().toISOString(),
} as ApprovalRecord,
],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Pending Approvals");
expect(prompt).toContain("Need Twitter API keys");
});
it("Manager prompt includes credential escalation guidance", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const ceo = shared.db.getAgent(ceoId);
const managerId = createAgent({ name: "CTO", role: "CTO", reportsTo: ceoId });
const icId = createAgent({ name: "Dev", role: "Engineer", reportsTo: managerId });
const ic = shared.db.getAgent(icId);
const ctx = buildCtx({
agentId: managerId,
directReports: [ic],
chainOfCommand: [ceo],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Break down work and delegate it");
});
it("IC prompt includes credential escalation guidance", () => {
const ceoId = createAgent({ name: "CEO", role: "CEO" });
const ceo = shared.db.getAgent(ceoId);
const icId = createAgent({ name: "Dev", role: "Engineer", reportsTo: ceoId });
const ctx = buildCtx({
agentId: icId,
directReports: [],
chainOfCommand: [ceo],
});
const prompt = buildSystemPrompt(ctx);
expect(prompt).toContain("Your manager is auto-notified and will help");
expect(prompt).toContain("explain what API key/account you need");
});
});