name	Opencode
description	OpenCode (open-source terminal coding agent by SST). Reads AGENTS.md natively. Rulebook generates opencode.json plus .opencode/commands, .opencode/agents, .opencode/skills.
version	1.0.0
category	cli
author	Rulebook
tags	["cli","opencode","agents","mcp"]
dependencies	[]
conflicts	[]

OpenCode Rules

Tool: OpenCode — open-source terminal coding agent.

Generated surface area

When Rulebook detects OpenCode (opencode.json / .opencode/ / opencode binary on PATH), it generates:

opencode.json with $schema, mcp.rulebook, and instructions. Existing user keys (model, theme, permission, custom agents) are preserved.
.opencode/commands/<name>.md for every user-invocable Rulebook slash command, with OpenCode-shaped frontmatter (description).
.opencode/agents/<role>.md for every Rulebook role agent (researcher, implementer, tester, code-reviewer, architect, docs-writer, security-reviewer, build-engineer, team-lead, etc.) with model tier mapped (haiku → claude-haiku-4-5, sonnet → claude-sonnet-4-6, opus → claude-opus-4-7) and a synthesized permission block.
.opencode/skills/<normalized-name>/SKILL.md for every Rulebook dev skill, name-normalized to [a-z0-9](-[a-z0-9])* (≤64 chars) and description bounded to ≤1024 chars.
.opencode/.rulebook-managed.json — sidecar listing managed keys so rulebook update knows what to refresh vs. preserve.

Idempotency

Files carrying the  marker are managed and refreshed on every rulebook update. Files without the marker are treated as user-owned and left untouched.

Sequential editing

OpenCode follows the same Rulebook discipline as Claude Code:

Edit files one at a time.
For 3+ files, decompose into 1–2 file sub-tasks.
Diagnostic-first: type-check / lint before tests.

Quality gates

Type-check / lint must be clean.
Tests must pass.
Coverage ≥95%.
Pre-commit hooks enforce these — never --no-verify.

MCP

mcp.rulebook in opencode.json exposes rulebook_task_*, rulebook_memory_*, rulebook_decision_*, rulebook_knowledge_*, rulebook_learn_*. Prefer MCP tools over shell commands when both exist.

Workflow

Reference AGENTS.md and .rulebook/specs/*.md.
Use /rulebook-task-create, /rulebook-task-list, /handoff.
Delegate to role agents (@researcher, @implementer, @tester).
Run quality checks before committing.
Save learnings (rulebook_memory_save, rulebook_knowledge_add).

Documentation

OpenCode Rules

Tool: OpenCode — open-source terminal coding agent.

Generated surface area

When Rulebook detects OpenCode (opencode.json / .opencode/ / opencode binary on PATH), it generates:

opencode.json with $schema, mcp.rulebook, and instructions. Existing user keys (model, theme, permission, custom agents) are preserved.

.opencode/commands/<name>.md for every user-invocable Rulebook slash command, with OpenCode-shaped frontmatter (description).

.opencode/agents/<role>.md for every Rulebook role agent (researcher, implementer, tester, code-reviewer, architect, docs-writer, security-reviewer, build-engineer, team-lead, etc.) with model tier mapped (haiku → claude-haiku-4-5, sonnet → claude-sonnet-4-6, opus → claude-opus-4-7) and a synthesized permission block.

.opencode/skills/<normalized-name>/SKILL.md for every Rulebook dev skill, name-normalized to [a-z0-9](-[a-z0-9])* (≤64 chars) and description bounded to ≤1024 chars.

.opencode/.rulebook-managed.json — sidecar listing managed keys so rulebook update knows what to refresh vs. preserve.

Idempotency

Files carrying the  marker are managed and refreshed on every rulebook update. Files without the marker are treated as user-owned and left untouched.

Sequential editing

OpenCode follows the same Rulebook discipline as Claude Code:

Edit files one at a time.

For 3+ files, decompose into 1–2 file sub-tasks.

Diagnostic-first: type-check / lint before tests.

Quality gates

Type-check / lint must be clean.

Tests must pass.

Coverage ≥95%.

Pre-commit hooks enforce these — never --no-verify.

MCP

Workflow

Reference AGENTS.md and .rulebook/specs/*.md.

Use /rulebook-task-create, /rulebook-task-list, /handoff.

Delegate to role agents (@researcher, @implementer, @tester).

Run quality checks before committing.

Save learnings (rulebook_memory_save, rulebook_knowledge_add).

Documentation