# orchestrate-omp

Adopt this skill: https://proskills.md/skills/other/dsebban-skills-skills-orchestrate-omp

Page: https://proskills.md/skills/other/dsebban-skills-skills-orchestrate-omp
This document is the public skill bundle for that page. It contains SKILL.md and every helper file the catalog carries. No login is required.

## File: skills/orchestrate-omp/SKILL.md
size: 6199
sha: 0523d346ee4e7d93bf34f4d10d3cbee512f3b6df
upstream: https://github.com/dsebban/skills/blob/main/skills/orchestrate-omp/SKILL.md

```text
---
name: orchestrate-omp
description: "Coordinate OMP subagents on substantial work. Use when parallel discovery, specialist review, or clearly partitioned implementation will shorten or strengthen the result; skip trivial tasks."
---

# Orchestrate on OMP

Stay available to the user while substantive work runs behind the scenes. The root coordinator owns the plan, user interaction, approvals, scope, integration, and final verification.

## 1. Frame

1. State the done predicate and the final artifact or decision.
2. Split only independent discovery, specialist review, or implementation with clear ownership. Keep tightly coupled work in one session.
3. Assign distinct primary ownership and name shared contracts before dispatch. Do not duplicate investigations or serialize a useful batch merely to avoid possible file overlap.
4. Keep external side effects and approval decisions with the root.

## 2. Choose agents

Use the exact names exposed by the live `task` tool; its inventory is authoritative. For the current OMP roster:

| Work | Agent | Effort when exposed |
|---|---|---|
| Narrow read-only codebase discovery | `scout` | `lo` |
| Routine scoped implementation | Omit `agent` for the default worker | `med` |
| Difficult or ambiguous implementation | Omit `agent` for the default worker | `hi` |
| Mechanical edits or data collection | `sonic` | `lo` |
| UI/UX implementation or review | Omit `agent` for the default worker; the brief carries the design specialization | `med` or `hi` |
| Code-quality verdict | `reviewer` | `med` or `hi` |
| Security verdict | `security-reviewer` | `hi` |
| External library or API research | `librarian` | `lo` or `med` |

When the live task schema exposes `effort`, match it to the assignment: `lo` for narrow questions, `med` for routine work, and `hi` when ambiguity or consequence justifies it. Otherwise rely on the selected agent's configured model role and do not pass an unavailable field. `/skill:setup-pstack` configures those roles.

Honor read-only and blocking markers in the live inventory. Never pass the default worker's name explicitly.

## 3. Dispatch

1. When the live schema exposes `tasks[]`, put independent participants in one batch so they start together and structure shared `context` as `Goal`, `Constraints`, and `Contract`. Otherwise start one participant per task call without passing `context`.
2. Give every item a stable CamelCase `name` of at most 32 characters, the most specific specialist `agent` when one fits, a complete standalone `task`, and a `solutionSpace` that states how open-ended the problem is rather than how much work it carries. Omit `agent` only for the default worker. Add `effort` only when the live schema exposes it.
3. Structure each assignment as `Target`, `Change`, and `Acceptance`. Include essential repository state, decisions, restrictions, and dependencies because children start without conversation history.
4. Give every writer an exact writable scope. When the live schema exposes `isolated`, use `isolated: true` for concurrent writers that need separate workspaces, then inspect the returned metadata to learn whether changes were applied or retained as artifacts.
5. Tell ordinary workers: `Complete this assignment directly. Do not call task or spawn subagents.` Allow nested delegation only when the assignment explicitly makes that worker a coordinator, and remember that `task.maxRecursionDepth` is 2 by default, so a coordinator's children are the last layer.
6. Tell workers to skip project-wide formatting, linting, and test suites. The root runs shared validation once after integration.

Large context belongs in a local file referenced with `local://<path>`, not duplicated across assignments.

Use `outputSchema` only when the coordinator needs a machine-readable result and the live schema exposes it. Set `schemaMode: "strict"` when invalid output must fail rather than return with a warning.

## 4. Coordinate

- Results auto-deliver. Continue handling user messages and independent root work instead of polling. `wait` is for the case where the root is genuinely blocked.
- `read proc://` lists live jobs and `read proc://<id>` inspects one without consuming its delivery.
- `write agent://<id>` sends a bounded correction, an answer, or the next coupled phase, and revives a parked session. `agent://all` broadcasts. Name expected dependencies in shared context so sibling messages are purposeful rather than discovery by negotiation.
- `write proc://<id>/kill` cancels stale or superseded work.
- Read complete results from `agent://<id>` and use `history://<id>` when a report is incomplete or suspicious.
- A completed job means the child yielded successfully. It does not mean its artifact is accepted.

## 5. Handoffs

A worker's final message is the handoff, and it is the only thing that moves information up the tree. Require the same shape from every worker: status (`PASS`, `ISSUES`, `BLOCKED`), branch or artifact paths, what it did, notes and deviations, suggested follow-ups.

- Keep the worker's words. Read `agent://<id>` for the full text rather than paraphrasing a truncated preview into the plan.
- `BLOCKED` is one task's dead end, not the tree's. Retry, repair, or clarify with a follow-up message and keep the rest moving.
- Stop the whole tree only when continuing would waste effort across every branch: a broken base, a wrong contract, a failing shared dependency.
- A child that dies without a report is a gap, not a success. Record it explicitly and decide whether to respawn.
- Treat the notes section as the richest part. Each deviation or concern is a candidate task, and complaints about unclear scope are evidence that the brief was too thin.

## 6. Integrate

1. Drain every required participant or record the cancellation or gap.
2. Inspect each claimed artifact and reconcile conflicts against the declared ownership and contracts.
3. Run the affected validation and real user path from the root at the integrated head.
4. Return one concise result with agent identifiers, accepted findings or changes, verification evidence, and unresolved gaps.

Claim model or backend diversity only when returned metadata proves it.
```

## File: README.md
upstream: https://github.com/dsebban/skills/blob/main/README.md

````text
# OMP skills

An OMP port of the [pstack](https://github.com/cursor/plugins/tree/main/pstack) skill catalog, plus the
OMP-native adapters that make it run on [`omp`](https://omp.sh/).

Upstream pstack targets Cursor. This repository tracks it and rewrites every host mechanism for OMP:
`task` batches instead of Cursor `Task` calls, agents instead of per-call model slugs, `skill://` instead of
relative paths, `agent://` and `history://` instead of cloud-agent URLs.

## Catalog

| Group | Skills |
|---|---|
| Router | `poteto-mode` routes substantial work to the matching playbook. Start here. |
| OMP adapters | `pstack-omp` maps canonical pstack roles and lifecycle protocols to OMP task agents. `orchestrate-omp` coordinates subagents on program-scale work. `setup-pstack` configures which model each role and agent runs on. |
| Understanding | `how`, `why`, `interrogate`, `blast-radius`, `recall`, `ketchup`, `in-your-own-words`, `teach` |
| Design and exploration | `architect`, `arena`, `swarm`, `figure-it-out`, `sharpen`, `reflect` |
| Execution | `tdd`, `correct`, `bro`, `show-me-your-work`, `benchmark-checklist`, `automate-me` |
| Writing | `unslop`, `technical-writing`, `no-comments` |
| Verification | `create-verification-skill`, `maintain-verification-skill`, `create-formal-verification-skill`, `maintain-formal-verification-skill` |
| Review | `fable-deslop` shrinks an oversized patch with the RepoPrompt CE context builder and its Oracle lanes |
| Language | `typescript-best-practices` |
| Bots | `make-bot-ui` |
| Help | `poteto-help` |
| Principles | 24 `principle-*` leaves cited by the playbooks |

`agents/` holds two optional subagent definitions, `poteto-agent` and `comment-sicko`, for imported
workflows that call those names directly.

## Requirements

- OMP 18.10.4 or newer.
- Bun, for the `poteto-mode` scripts.
- `gh`, for the PR playbooks.
- RepoPrompt CE and its `rpce-cli`, for `fable-deslop`.
- `trash`, for the worktree-cleanup playbook.

## Install

Clone the repository and point OMP at its `skills/` directory:

```bash
git clone https://github.com/dsebban/skills.git ~/dev/skills
```

```yaml
# ~/.omp/agent/config.yml
skills:
  customDirectories:
    - ~/dev/skills/skills
```

The clone stays live, so `git pull` updates every skill at once.

Install the optional agents by copying them into OMP's user agent directory:

```bash
mkdir -p ~/.omp/agent/agents
cp ~/dev/skills/agents/*.md ~/.omp/agent/agents/
```

Start a new OMP process after either change. Existing sessions keep their previous inventory.

## Configure models

OMP routes models by role, not per call. Run this once:

```text
/skill:setup-pstack
```

It detects the available catalog, asks for a reasoning budget, and writes `modelRoles` and
`task.agentModelOverrides` into the active profile's `config.yml` after validating every selector. It also
reports when two panel agents resolve to the same model, which is the difference between a second opinion
and self-review.

## Use

Start with `/skill:poteto-mode` and describe the work; it routes to the matching playbook and pulls in the
skills that playbook needs. Every skill is also directly invocable as `/skill:<name>`.

Upstream marks the catalog `disable-model-invocation: true`, and this port keeps that. The model does not
pick these skills up on its own; the user invokes them, or `poteto-mode` routes to them. Three skills are
the exceptions and load when the work calls for them: the OMP adapters `pstack-omp` and `orchestrate-omp`,
and `setup-pstack`, whose trigger phrases are about configuring models.

Reference files resolve through the owning skill:

```text
skill://poteto-mode
skill://poteto-mode/playbooks/babysit.md
skill://why/references/epistemics.md
```

OMP resolves skill names exactly and case-sensitively, and `skill://` paths are readable but not executable.
A step that runs a bundled script resolves its real path first with `realpath skill://<name>`, which prints
the physical directory under any install root.

`/skill:poteto-help` explains the catalog, including which upstream surfaces have no OMP equivalent.

## OMP task behavior

- Use the exact agent names in the live `task` inventory. Omit `agent` to select OMP's default worker.
- Use `tasks[]`, `context`, `effort`, and `isolated` only when the live task schema exposes them.
- Never put a model in a task item. Model choice is `modelRoles` and `task.agentModelOverrides`.
- Treat child sessions as fresh context. Put the goal, constraints, ownership, acceptance criteria, and the
  leaf-agent boundary in every assignment.
- Read complete results from `agent://<id>` and transcripts from `history://<id>`.
- Keep user interaction, approvals, integration, and final verification with the root coordinator.

## Provenance

- The catalog is adapted from Lauren Tan's [`pstack`](https://github.com/cursor/plugins/tree/main/pstack),
  version 0.15.28 (`cursor/plugins` at `d457567`).
- `orchestrate-omp` began as Eric Provencher's
  [`orchestrate`](https://github.com/provencher/codex-skills/tree/main/orchestrate) and is now OMP-native.
- `pstack-omp`, `fable-deslop`, and the OMP integration, including `setup-pstack`'s model helper, are
  maintained here.

See [`THIRD_PARTY_NOTICES.md`](THIRD_PARTY_NOTICES.md) for upstream copyright notices.

## License

MIT. See [`LICENSE`](LICENSE).
````
