omp-delegate — скилл для Codex: делегирование задач в omp

---
name: omp-delegate
description: >-
  Delegate a coding task to Oh My Pi (`omp`) as a background implementer, then review its
  diff and land it yourself. Use this whenever the user wants to delegate implementation work to
  Oh My Pi / omp - phrasings like "have omp implement X", "delegate this to oh my pi", "run it
  through omp", "use oh-my-pi to implement/fix/refactor" - or wants to run a queue of coding
  tasks through omp while staying the reviewer. DO NOT USE for tasks small enough to do inline,
  when the user wants the code written directly without delegating, or when they mean the original
  Pi CLI (`pi`) — that is pi-delegate.
license: MIT
compatibility: Requires the `omp` CLI installed and authenticated (`/login` inside omp, or a provider API-key environment variable), Node 18+, and git. The orchestrating agent must be able to run shell commands and read files. Shell examples assume bash/zsh (macOS/Linux, or Git Bash/WSL on Windows).
metadata:
  version: 0.5.0
---

# Oh My Pi Delegate

You are the **orchestrator**. Delegate a bounded coding task to a separate **implementer** - Oh My
Pi (`omp`) - then review what it produced and land it yourself. You write the brief and own the
judgment; the implementer makes changes in its own session; you verify and commit.

The loop needs only a shell command and file access, so any comparable orchestrator can drive it.

## The binary is `omp`, not `pi`

Oh My Pi is a fork of Pi. This skill drives **`omp`** (`@oh-my-pi/pi-coding-agent`). The original
Pi CLI is a different binary (`pi`) with a different skill (`pi-delegate`). If `omp` is missing but
`pi` is installed, you have Pi, not Oh My Pi.

## When NOT to use this

- The task is small enough to do inline; delegation overhead is not worth it.
- The `omp` CLI is not installed or authenticated.
- The user asked for the original Pi CLI (`pi`) — use `pi-delegate`.
- You need a sandboxed implementer. Oh My Pi has no sandbox. `--read-only` restricts the tool
  surface; a write-capable run executes without prompts (`--yolo`).

## Prerequisites (check once)

1. Install omp with `bun install -g @oh-my-pi/pi-coding-agent` (or the install path from
   https://omp.sh).
2. Authenticate: `/login` inside omp for a subscription provider, or an API-key environment
   variable for an API-key provider. Credentials live under `~/.omp/`.
3. Confirm `omp --version` succeeds.
4. Work in, or point `--cd` at, the target git repository.

## Choose the model (optional)

Omit `--model` (and `--provider`) to use omp's configured default for this project / profile. The
catalog is **this install's** authenticated providers — not a fixed list in this skill.

To pick another model:

1. **List what this install can actually run.** Do **not** pass `omp --list-models` — that flag is
   gone and omp treats it as an unknown flag (exit 2). Use the `models` subcommand:
   - `omp models` — every available model, grouped by provider
   - `omp models --json` — the same catalog, machine-readable
   - `omp models find <substring>` — filter by provider, id, or name (example:
     `omp models find sonnet`)
   - `omp models <provider>` — one provider's models
2. **Pass that id to the relay.** `--model <pattern>` is omp's own `--model`: a fuzzy match against
   the catalog (provider/id, a bare id, or a unique substring). `--provider <name>` pins the
   provider when the pattern is ambiguous.
3. The relay forwards only letters, digits, and `. _ : / -`. Glob patterns with `*` are rejected.

`--thinking <level>` is a separate reasoning dial, not a model id. Allowed values: `off`, `auto`,
`minimal`, `low`, `medium`, `high`, `xhigh`, `max`. The relay rejects anything else (including
`inherit`) before dispatch — omp would otherwise warn and ignore a bad value.

A fleet lane (`--lane`) can set `provider`, `model`, and `effort`. Lane `effort` becomes
`--thinking`; an explicit `--thinking` / `--model` / `--provider` flag wins over the lane.

The relay does not forward `--api-key`, `--smol`, `--slow`, or `--plan`. Those stay omp's own CLI.

## The loop

Run these five steps per task. Steps 1, 4, and 5 require judgment; 2 and 3 are mechanical.

### 1. Write the brief

Oh My Pi sees only the text you send plus what it can inspect in the workspace - no chat history or
shared context. Include the goal, current state, what to change, what to leave untouched, the
project's **actual** gates, and a report contract. Tell omp not to commit. Keep one task per brief.
omp auto-loads `AGENTS.md`/`CLAUDE.md` context files from the workspace and its parents, so repo
instructions reach it without inlining. See
[references/writing-the-brief.md](references/writing-the-brief.md).

### 2. Dispatch

For **Codex orchestration of background work**, install and configure the separate
**codex-background** support utility first. If its MCP tools or registry are missing, stop and
report the setup requirement; do not silently fall back to model-driven polling. Then invoke its
registered `delegate_run` MCP tool once in place of the shell example below.
Follow that skill for structured common arguments, unchanged provider flags and reattachment by
runId. Keep the request pending until completion, then continue this skill's review and landing
steps. Other orchestrators keep the shell workflow below.

Use the bundled relay. It pipes the brief to `omp --mode json` on stdin, captures the JSON event
stream, and writes `result.json`. (`<skill-dir>` is the installed folder containing this
`SKILL.md`.)

```bash
node "<skill-dir>/scripts/relay.mjs" --brief brief.txt --cd /path/to/repo
# list models first:                       omp models   (or: omp models --json)
# choose a model:                          add --model <id from omp models>
# choose a provider:                       add --provider <name>
# set thinking level:                      add --thinking high
# read-only run (review/diagnosis):        add --read-only
# trust project .omp resources:            add --approve
# resume the most recent session:          add --resume-last  (delta brief only)
# resume a specific session:               add --session <id> (delta brief only)
# hard time limit (watchdog):              add --timeout 2h  (the 30m default suits short runs; implementation briefs routinely need 1-2h)
# see all options:                         node .../relay.mjs --help
```

The child process's cwd pins the workspace. The relay writes artifacts under the system temp dir
by default and never commits. See [references/dispatch-and-poll.md](references/dispatch-and-poll.md).

### 3. Wait for completion

Codex using codex-background keeps the original MCP request pending: no model-driven sleep,
status or log polling. Use `delegate_wait` only to reattach after an interrupted request.

The relay blocks until omp finishes. Run it with the orchestrator's background-command facility,
or background it in the shell and poll for `result.json`. A pre-run usage error exits 2 and writes
no result; a missing `omp` exits 127 and writes `status: "omp_unavailable"`.

Trust process state and the working tree over a progress display. Completion means the process
exited and `result.json` exists. omp's full report is the `finalMessage` field in `result.json`
(also printed in full on stdout between the report markers).

### 4. Review - do not trust the self-report

Treat omp's final message and gate claims as claims:

- Re-run the project's gates yourself.
- Read the diff against the brief, starting with `touchedFiles`.
- Run relevant guard skills if installed.
- Round-trip migrations and grep for dangling references after removals or renames.

See [references/review-and-land.md](references/review-and-land.md).

### 5. Land it

The implementer edits the working tree; **the orchestrator commits.** Commit only after the gates
pass and the diff holds. If rework is needed, send a delta brief with `--resume-last` or
`--session <id>`, then review again.

## Autonomy and permissions

Oh My Pi has **no sandbox**. Print mode has no approval UI, so a write-capable relay run always
passes `--yolo` (`tools.approvalMode: yolo`) — otherwise a user's `always-ask` or `write` config
would stall until the watchdog. The other controls are:

1. `--read-only` restricts omp's callable tools to `--tools read,grep,glob`. It does not pass
   `--yolo`. Installed extension code still runs with the user's host permissions if project
   resources are trusted.
2. The relay passes `--no-extensions --no-skills --no-rules` by default, so project `.omp`
   extensions, skills, and rules stay undiscovered. `--approve` is the explicit opt-in for a
   repository the user trusts.
3. `touchedFiles` and the diff are the record of what changed. Inspect them after every run.

## Authorization model

Delegation is something the human opts into. Once they have ("run this queue", "proceed"),
committing verified, gate-passing work is the agreed contract. Two limits remain: **surface, don't
absorb** (report omp's design decisions, defensible-but-unasked turns, and non-blocking nitpicks)
and **stop for scope changes** (if correct completion needs going beyond the brief, ask instead of
expanding the mandate). See [references/review-and-land.md](references/review-and-land.md).

## References

- [references/writing-the-brief.md](references/writing-the-brief.md) - structure, report contract,
  real gates, stdin delivery, model listing, and delta briefs.
- [references/dispatch-and-poll.md](references/dispatch-and-poll.md) - flags, artifacts,
  `result.json`, polling, and failure recovery.
- [references/review-and-land.md](references/review-and-land.md) - review checklist, commit
  boundary, and rework through omp sessions.
- [references/multi-task-queues.md](references/multi-task-queues.md) - sequential queues,
  constraint carry-forward, progress tracking, and the final coherence pass.