Customization

Project Context

Create or update AGENTS.md (created on first run or via /init) with a short project summary:

# Project

## Tech
- Node.js, TypeScript, PostgreSQL

## Standards
- Async/await everywhere
- Tests for new features
- No hardcoded secrets

Agent Configuration

Edit .opencode/agent/[agent].md to adjust behavior (model, permissions, tools).

Custom Commands

Add command files to .opencode/commands/ with frontmatter (description, agent). Type /command-name in the TUI to run.

Skills

Skills live in .opencode/skills/ and can be loaded on demand:

• project-bootstrap
• agent-diagnostics
• docs-validation

Recommended policy:

• Load skills on demand only when the task clearly matches the domain.
• Use one relevant skill by default; add a second only for explicit cross-domain work.

Skill Permissions per Agent

Harden each agent with permission.skill allowlists:

permission:
  skill:
    "*": "deny"
    "blogger": "allow"
    "brutal-critic": "allow"

This prevents unrelated skills from being loaded and keeps behavior predictable.

Scope Policy (Core-only)

Current distribution is intentionally core-only; no optional skill packs are shipped.

When evaluating potential additions, use this gate:

1. Repeated demand in real usage
2. Clear capability gap vs current core
3. Named maintenance owner
4. Clean licensing/provenance (or clean-room rewrite)

Next Steps

• Getting Started
• Agents
• Coding Standards