| name | customize-opencode |
| description | Use ONLY when editing or creating opencode's own config: opencode.json, skills, commands, agents, plugins, MCP servers, instructions, permissions, security rules. Not for application code. Trigger on 'configure opencode', 'opencode setup', 'add skill to opencode', 'update opencode config', 'opencode plugin'. |
Customize opencode
Configuring opencode itself. Not for application code.
Workflow
- Understand — read existing config, check vault for current setup
- Propose — brief plan with files to change, wait for confirmation
- Apply — surgical edits, validate JSON
- Update vault — always update
setup.md + relevant files in CODE/LLMs/
- Verify — check validity, tell user to restart if needed
Skill Composition
- Before reading or editing vault documentation, load and apply
obsidian-vault for the canonical vault path, frontmatter preservation, and filesystem safety rules.
Vault sync rules
| Change | Vault file |
|---|
| Any config change | CODE/LLMs/opencode/setup.md |
| New skill | CODE/LLMs/Skills/<name>.md |
| New command | CODE/LLMs/Commands/<name>.md |
| New agent | CODE/LLMs/Agents/<name>.md |
| New instruction | CODE/LLMs/Instructions/<name>.md |
| Plugin change | CODE/LLMs/Plugins/<name>.md |
| Model change | CODE/LLMs/LLMs.md |
| MCP change | CODE/LLMs/MCP/<name>.md |
Never document
API keys, tokens, auth cookies, workspace IDs, private keys, OAuth secrets.
Config reference
| File | Purpose |
|---|
opencode.json | Main config |
AGENTS.md | Global agent instructions |
security-rules.md | Credential + injection protection |
opencode-quota/quota-toast.json | Quota UI |
skills/<name>/SKILL.md | Local skills |
commands/<name>.md | Slash commands |
agents/<name>.md | Agent definitions |