AGENTS.md

1# .opencode/ — Project-Scope Skills & Commands
2 
3**Generated:** 2026-05-20
4 
5## OVERVIEW
6 
7Project-scope OpenCode configuration: 5 skills and 5 slash commands committed alongside the source. Picked up by [`packages/omo-opencode/src/features/opencode-skill-loader/`](../packages/omo-opencode/src/features/opencode-skill-loader/) and the slash-command discovery pipeline.
8 
9**Relationship to `.agents/`:** `.agents/` is the migration target during the `oh-my-opencode` → `oh-my-openagent` rename. It is a SUPERSET of `.opencode/` (mirrors all 5 skills + adds 5 more, mirrors the 5 commands). Both directories load during the transition; consumers should prefer `.agents/`.
10 
11## SKILLS (5)
12 
13| Skill | Purpose |
14|-------|---------|
15| `work-with-pr/` | Full PR lifecycle skill: worktree → implement → atomic commits → PR → verification loop → merge. |
16| `work-with-pr-workspace/` | Iteration workspace for `work-with-pr` — captures benchmark inputs for the iteration-* subdirs. |
17| `github-triage/` | Read-only triage of issues + PRs. 1 item = 1 background task. Writes evidence-backed reports. Never takes GitHub actions. |
18| `hyperplan/` | Adversarial multi-agent planning. 5 hostile category members cross-critique, lead synthesizes. |
19| `pre-publish-review/` | Nuclear-grade 16-agent pre-publish release gate. Runs `/get-unpublished-changes`, spawns ultrabrain agents per change, oracle for synthesis. |
20 
21Each skill follows the standard layout (`SKILL.md` + optional `scripts/`, `references/`, `assets/`).
22 
23## COMMANDS (5 slash commands)
24 
25| Command | Purpose |
26|---------|---------|
27| `/get-unpublished-changes` | Compare HEAD with the latest published npm version and list unpublished changes. |
28| `/omomomo` | Easter-egg command — about oh-my-opencode. |
29| `/publish` | Publish via GitHub Actions workflow. Argument: `<patch|minor|major>`. |
30| `/remove-deadcode` | Remove unused code with LSP-verified safety + atomic commits. |
31| `/security-research` | Run the Team Mode security-research audit with 3 vulnerability hunters and 2 PoC engineers. |
32 
33## OTHER CONTENTS
34 
35- `background-tasks.json` — Runtime state for active background tasks (gitignored from sensitive content but tracked structurally).
36- `bun.lock`, `package.json`, `node_modules/` — Skill dependencies (some skills install their own packages).
37- `.gitignore` — Local scope ignore patterns.
38 
39## CONVENTIONS
40 
41- **Skill YAML frontmatter is mandatory.** [`opencode-skill-loader`](../packages/omo-opencode/src/features/opencode-skill-loader/) rejects skills without `name` + `description`.
42- **Project-scope > user-scope.** A skill at `.opencode/skills/X/` overrides `~/.config/opencode/skills/X/` of the same name.
43- **Trigger words** in the skill description determine when OpenCode loads the skill. Be specific.
44- **Commands are user-visible.** Name them with a leading `/` (the loader normalizes filename → command).
45 
46## ANTI-PATTERNS
47 
48- Never duplicate a built-in skill from `packages/omo-opencode/src/features/builtin-skills/` here without a clear reason.
49- Never commit `background-tasks.json` contents that include session IDs or token secrets.
50- Never write a skill that calls `session.promptAsync` directly — go through `dispatchInternalPrompt` (see root AGENTS.md "Internal message injection is dangerous").
51