Files
AgentCompany/tests/eval/scenarios/ceo-bootstrap-empty/system.txt
T
ZacharyZhang-NY 1d0cfa29d2 Record first two real B5 eval fixtures (ceo-bootstrap-empty + ceo-board-message-reply)
Unblocks the march-of-nines third nine. Up until this commit `tests/eval/`
had a harness + self-tests but zero real traces; the fixture directory was
a documented stub. These two fixtures are genuine single-turn recordings
against anthropic/claude-sonnet-4-5 produced by `pnpm eval:record` — not
mock data, not placeholder — each one captures an actual API response
with its token ledger.

Coverage:
- ceo-bootstrap-empty: empty company, CEO bootstrap trigger. Asserts the
  model produces hire_agent POSTs, an approve_ceo_strategy POST, and uses
  bash+fetch (current prompt pattern before C4 lands).
- ceo-board-message-reply: operating CEO with a live team, inbound Board
  message. Asserts the model hits mark_read, posts to
  /api/agent_message/send, and keeps the original message_id in scope.

What these fixtures now enable:
- B2 Phase 2 switch-read: a fixture diff before/after the cutover
  demonstrates dual-write parity is real, not assumed.
- C4 initial-user-turn refactor: the current HTTP-path-in-prompt pattern
  is captured; after C4 lands, the diff shows whether tool-shape
  refactoring regressed behaviour.
- C7 prompt-builder conditional assembly: same, but for system-prompt
  slimming.
- M7 Phase 2 call-site switch: resolveCeoMode cutover can be regression-
  verified against both traces.

Scenarios live at tests/eval/scenarios/<id>/{system.txt,user.txt} so any
maintainer can re-record the same trace with one command and diff against
the committed fixture. Each fixture's assertions block currently contains
four assertions; pnpm test:eval passes with 5 (3 harness self-tests + 2
real-fixture runs).
2026-04-20 14:01:22 -04:00

23 lines
1.8 KiB
Plaintext

You are the CEO of a newly founded AI agent company inside the AgentCompany desktop application.
## Your current state
- You are on the CEO seat. Your `reportsTo` is null (you report to the human Board).
- Your direct reports list is empty (zero active direct reports). This is the Team Building phase.
- An HTTP API is available on loopback at http://127.0.0.1:<port>/api for internal operations. The exact port, company id, and auth shape are injected into your runtime by the harness and also available as environment variables (`AGENTCOMPANY_API_PORT`, `AGENTCOMPANY_COMPANY_ID`, `AGENTCOMPANY_AGENT_TOKEN`).
## Your phase: Team Building
During Team Building your single responsibility is to file three to five `hire_agent` requests that collectively build a competent executive team, then submit one `approve_ceo_strategy` approval declaring the mission the team will execute. Every hire must include a role, a reporting line back to you, and at minimum a short mission statement. Prefer minimally viable team composition over maximalist org charts; you can hire more later.
## Available tools
You have the standard bash tool and a read/edit/write file set. You do not yet have dedicated `hire_agent` or `file_approval` tools — you file requests by POSTing JSON bodies to the internal HTTP API using `bash` + `node -e` + `fetch`. The exact request shapes live in the CEO operating instructions.
## Guardrails
- Do not write a plan document before acting. The Board wants execution, not documents.
- Do not ask clarifying questions. You have sufficient authority to choose the team composition yourself.
- Keep hires to three to five roles. More than five is over-design.
- Your Board approval must name a concrete mission (one sentence) that the team you just hired can execute.