@haylan/opencode-pocock

OpenCode v2 integration for Matt Pocock's skills: the upstream skills injected verbatim as OpenCode skills, plus generated agents that run his skill-driven engineering flow — grill an idea, turn it into a spec and tickets, build test-first with parallel workers in isolated git worktrees, review, retro.

All engineering discipline is Matt Pocock's; this package only supplies the OpenCode v2 wiring. Upstream is synced nightly and the package republished on change (see CHANGELOG.md for the upstream revision each release carries).

Install

Add the plugin (from the Gitea registry, or npm once published there):
```
// opencode.json
{
  "$schema": "https://opencode.ai/config.json",
  "plugins": ["@haylan/opencode-pocock"]
}
```
The plugin injects all skills at startup — no files copied into your repo.

Install the agents (OpenCode discovers agents only as markdown files, and plugins cannot register them, so this one copy step is required):

npx @haylan/opencode-pocock init      # project: .opencode/agents/
npx @haylan/opencode-pocock init -g   # or global: ~/.config/opencode/agents/

Once per repo, run /setup-matt-pocock-skills in OpenCode. It configures the issue tracker, triage labels and domain docs that to-spec, to-tickets, triage, implement, implement-spec, code-review and wayfinder depend on.

Usage

Start OpenCode with the pocock agent selected and describe what you want to build:

opencode --agent pocock

Agent	Role
`pocock`	Primary orchestrator. Routes with `ask-matt`, walks grill → spec → tickets → build → review → retro, proposes before each user-invoked step, runs `implement-spec` with one git worktree per ticket and parallel `pocock-worker` subagents.
`pocock-worker`	Subagent. Builds one ticket test-first (`tdd` at the spec's seams) inside its worktree, merges the integration tip, reports. Never pushes.
`pocock-reviewer`	Subagent. Read-only executor of one `code-review` axis (Standards or Spec).

23 of the 31 vendored skills are user-invoked upstream (disable-model-invocation: true); OpenCode v2 honors that natively — they stay out of the model's auto-selection and are run by you as slash commands (/to-spec, /retro, ...).

Both agents may create worktrees under /tmp/pocock-workers. If you use subagents that write there (exploration notes), consider a global permission rule:

{ "permission": { "external_directory": { "/tmp/pocock-workers/*": "allow" } } }

Plugin options

{
  "plugins": [
    {
      "package": "@haylan/opencode-pocock",
      // Ports skills/misc/git-guardrails-claude-code to OpenCode: blocks dangerous git
      // commands (push --force, reset --hard, clean -f, branch -D, ...) before execution.
      "options": { "guardrails": true }
    }
  ]
}

Development

npm run sync — update upstream/ snapshot from mattpocock/skills (script-only; never hand-edit upstream/ or skills/).
npm run generate — regenerate skills/ + manifest.json from upstream/ and .opencode/state/classification.json.
npm test / npm run check / npm run build — unit + integration tests (node --test on TS), prettier + tsc gate, build.
npm run smoke — boots opencode serve against a fixture and verifies plugin/skills/agents over the HTTP API (needs a local opencode binary; no model calls).

Design and rationale: docs/architecture.md. Per-skill classification: docs/skill-mapping.md.

License

MIT — see LICENSE and NOTICE for upstream attributions.

3.7 KiB Raw Blame History