メニュー
Run the forge gate-chain commit workflow: classify the change, run category validation, changelog check, and adversarial review before any git commit. Use whenever committing changes in this repo.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Install and initialize the forge agent operating system (DVRR worktrees, gate chain, adversarial review, evals) in the current repo. Use when asked to run forge-init, install forge, or initialize the forge system.
Merge the current forge worktree back to the default branch through the 4-gate verification chain (tests, lint, adversarial review, summary) and a locked rebase. Use when finishing a worktree session or asked to merge work back.
Reviews a GitHub issue to determine validity, classify as user error or bug, check if already fixed, and take appropriate action. For user errors, improves error messages and documentation. For real bugs, implements a fix following the full commit workflow including tests and adversarial review. Closes the issue with a clear resolution message. Use when the user says "review issue", "triage issue", "is this a bug", "check issue", "issue review", or provides a GitHub issue URL or number.
Reviews all open pull requests on the project's GitHub repository and produces a structured report. Classifies PRs as mergeable, needs-work, stale/out-of-date, or duplicate. Use when the user says "review PRs", "PR report", "check pull requests", "PR status", "open PRs", or "duplicate PRs".
Deep adversarial code review using the 8-lens review constitution. Examines diffs for correctness, security, completeness, and project-specific concerns. Use when performing pre-commit reviews, quality-loop iterations, or on-demand code audits. Loaded by review-cheap and review-final agents.
Deep adversarial specification review using the 8-lens review constitution. Evaluates design documents, plans, and specs for ambiguity, completeness, feasibility, security, and project-specific concerns. Loaded by review-cheap and review-final agents.
| name | commit |
| description | Run the forge gate-chain commit workflow: classify the change, run category validation, changelog check, and adversarial review before any git commit. Use whenever committing changes in this repo. |
Follow the COMPLETE pre-commit workflow defined in .opencode/rules/commit-workflow.md. Do NOT skip any steps unless the user explicitly says "skip tests", "skip review", or "just commit".
Validation is mandatory and must be executable where possible. Do not rely only on static inspection.
Parallel session safety is mandatory: commit only files changed in the current session. Never include files you did not edit in this session, and never use blanket staging (git add . / git add -A).
The authoritative steps are in commit-workflow.md; do not re-derive or skip any:
git status --short; identify only files YOU changed this session; classify each by category per [[commit-workflow]] Step 1. Files under .opencode/rules/**, .opencode/agents/**, .claude/agents/**, .opencode/commands/**, .claude/commands/**, .opencode/skills/**, .opencode/plugins/**, .opencode/scripts/**, opencode.jsonc, .claude/settings*.json, the review constitution, or CI/test gates are the control (AI-component) class — higher-scrutiny (see step 4).bash .opencode/evals/run-evals.sh (a regression blocks).review-cheap subagent (fresh context) on the diff; must return PASS. Control / AI-component changes use the authoritative review-final and are gated-approval — present the PASS to the user and get explicit approval before committing (separation of duties; never auto-commit a control change). Cap the review loop at 8 iterations; if it does not converge, record the residual risk and escalate — do not commit..opencode/scripts/check-halt.sh commit (if it exits non-zero, stop — an operator halt is in force); then stage files by explicit path (NEVER git add .) and commit with a descriptive message.If the user provided additional instructions: $ARGUMENTS