Files
codex-crew/skills/codex-agent/SKILL.md
T
admin 54aabcab23 fix: review mode works with prompts, anti-self-work contract, EN docs
- codex-run.sh review: CLI forbids scope flags + prompt (stdin '-' counts
  as [PROMPT]); translate scope into prompt text, native review only when
  prompt.md is empty. Add --skip-git-repo-check for consistency.
- agents/codex.md: marker contract (=== TASK FOR CODEX === / === END TASK ===,
  legacy RU markers accepted) so wrapper instructions never leak into the
  codex prompt and the wrapper never mistakes the task for its own job;
  self-check against doing the task itself; preamble now states codex is a
  subagent executor and process flows belong to the orchestrator; scope
  flags as separate args; no nohup; Read before rewriting prompt.md.
- skills: wrapper framing line + marker block in prompt templates.
- All instructions and README translated to English (RU trigger phrases
  kept in skill descriptions).
- tests: 3 new review-mode cases, ALL PASS.
- bump 0.1.2.
2026-07-28 19:12:19 +05:00

50 lines
2.2 KiB
Markdown

---
name: codex-agent
description: Delegate execution to Codex — write code, fix a bug, research. Use when user says «отдай кодексу», «пусть codex сделает/напишет/починит», «делегируй в codex», "delegate to codex", or when a plan step is assigned to Codex as the implementer.
---
# Delegating a task to Codex
Spawn the `codex` subagent (Agent tool). It keeps a live codex session:
refinements and answers to its questions go via SendMessage to the same
agent, NOT a new spawn.
## Subagent prompt — required parts
- First line — framing for the wrapper: "You are a wrapper around the codex
CLI: do not solve the task yourself, launch codex and return its answer +
thread_id". Without it the wrapper may take the task as its own and do it
instead of codex.
- `cwd: <absolute project path>`
- The task — in a block between markers `=== TASK FOR CODEX ===` and
`=== END TASK ===`: the wrapper forwards only this block to codex, so your
wrapper instructions don't leak into the prompt. Inside the block:
- Task: what to do. Concrete, with relevant files/paths.
- Acceptance criteria: what must work, which tests/commands must pass.
- What to return: report, list of touched files, how verified.
- Optional: `codex options: -m o3 -s read-only ...`
## Example
```
You are a wrapper around the codex CLI: do not solve the task yourself, launch codex and return its answer + thread_id.
cwd: /abs/path/to/project
=== TASK FOR CODEX ===
Task: fix the N+1 in GET /orders — the queries log shows one query per order
line item. See app/controllers/OrdersController.php.
Acceptance: existing tests pass (composer test), ≤3 DB queries per request.
Return: list of changed files, the essence of the fix, composer test output.
=== END TASK ===
```
## After completion
- Codex answer = report + `thread_id: ...`. Don't take it on faith: check
the diff, run the tests yourself.
- Remarks → SendMessage to the same subagent (same codex session, context
intact).
- Codex ended with a question (blocker) → answer via SendMessage, it
continues.
- While it's working you can ask for status via SendMessage — doesn't
interrupt.