Skip to content

donaldfilimon/abi / .agents/skills/opencode

opencode

Verified

Set up and operate abi through opencode — how opencode discovers abi skills, slash-commands, and MCP servers. Use when the user asks about opencode config, why an abi skill is not loading in opencode, or how to add a skill to the opencode allowlist.

FileKindSize
SKILL.mdskill md2.6 KB
---
name: opencode
description: Set up and operate abi through opencode — how opencode discovers abi skills, slash-commands, and MCP servers. Use when the user asks about opencode config, why an abi skill is not loading in opencode, or how to add a skill to the opencode allowlist.
---

# opencode

opencode loads abi via mechanisms wired in repo-root `opencode.json` and the
`.opencode/skills` symlink:

- **Skills**: `.opencode/skills` is a symlink to `../.agents/skills`, so every
  canonical skill dir is visible to opencode automatically. No separate allowlist
  file is required today — editing a skill under `.agents/skills/<name>/` is enough.
- **Agent TUI slash commands**: live in the Rust REPL, not an OpenCode JSON
  map. The current surface is `/help`, `/model`, `/profile`, `/sea`, `/status`,
  `/context`, `/history`, `/reset`, `/features`, `/clear`, and `/quit` plus
  documented aliases. Workflow skills for Git or session planning do not add
  `/diff`, `/commit`, `/save`, or `/load` to the binary.
- **MCP servers**: standardized core set across CLIs (post-2026-09 cleanup):
  `abi-mcp` (via full-path or `./mcp/launcher.sh stdio`),
  `skill-loop` (@0.3.3 pinned),
  `context7`, `fetch` (uvx), `filesystem` (~/.abi/ with ~/.abi/** allow),
  `github` (unauth ok), `memory`, `playwright`.
  Global: `~/.config/opencode/opencode.jsonc` + `~/.cursor/mcp.json` + `~/.claude.json` + `~/.codex/config.toml` + `~/.grok/config.toml`.
  Project (abi/): `.mcp.json`, `.cursor/mcp.json`, `opencode.json`, `.codex/config.toml` use relative launcher + minimal (abi-mcp + skill-loop).
  Comments present for key-requiring servers; enabled=false for brave/postgres/sentry/supabase.
- **Instructions**: `opencode.json` `instructions` points at `AGENTS.md`,
  `tasks/lessons.md`, and `tasks/todo.md`.

## To add a skill for opencode
1. Author/complete the skill under `.agents/skills/<name>/SKILL.md`.
2. Run `tools/check_skills.sh <name>` to confirm frontmatter is valid.
3. Skills auto-reload from the symlink — no restart needed for content edits.

## Gotchas
- `.opencode/skills` is a symlink; editing a skill there edits the canonical copy.
- `skill-loop` writes `.skill-telemetry/registry.json` plus amendment proposals
  under `.skill-telemetry/` automatically (content-drift proposals). Do not
  hand-edit the registry. (`.skill-telemetry/` is gitignored.)
- The `.agents/skills/sync-clis/launch.sh` syncer updates in-repo
  `.claude/skills/` and `.grok/` from the canonical tree (creates missing skill
  dirs; skip-list skills are not copied). opencode needs no sync (symlink), but
  Claude Code / grok do.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall