Have Codex implement a concrete work order and have Claude independently inspect the final changes. Preserves the explicit Codex builder choice; use claudex-loop for host-based planning and automatic reviewer selection.
インストール方法を見るインストールする前に、エージェントに与えられる指示の中身を確認できます。
Compatibility entry point: the builder is explicitly Codex. The shared workflow and executable live in the sibling claudex-loop skill; install that skill alongside this one.
Load the shared build reference and runtime reference. Preserve SPEC_FILE (mapped to the runner's --plan), LOG_FILE, PROOF_CMD, MAX_FIX_ROUNDS, and explicit model/effort arguments. The spec may have any filename; never substitute PLAN.md silently.
build --host claude --builder codex path. Claude then inspects all changes and independently runs proof checks. Log evidence and limitations. A fresh Claude CLI inspector is also available through the shared runner if useful.inspect --host codex --builder codex. Do not ask Codex to certify its own changes as independent review.Use the current valid plan approval when this follows claudex-loop. An explicitly requested standalone work order can use --unreviewed-spec, with that status recorded. If the spec still needs consequential decisions, settle those before delegating; do not build by inventing missing requirements.
Preserve unrelated changes, record the baseline, independently run the proof, inspect staged and new files as well as the unstaged diff, and keep fix rounds bounded. After a Claude takeover, require fresh Codex inspection of Claude's edits; if both contributed, log authorship and have each inspect the other's changes. Never present an earlier review as covering later fixes. Present the completed diff for any remaining sign-off required by the user's authorization.
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Harden a plan with independent Claude/Codex review, then optionally build and cross-inspect it. Start in either Claude Code or Codex: the host plans and the other provider reviews. Use for claudex this plan, claudex-loop, or the legacy crucible trigger; not for trivial edits.
日本語の概要は準備中です。原文の説明を表示しています。
Recommend a model and a scoped handoff for a task in Claude Code or Codex. Use when choosing who should handle a task, seeking a second opinion, getting unstuck, or delegating focused work; execute one handoff when requested.
日本語の概要は準備中です。原文の説明を表示しています。
Have Codex independently review an existing implementation plan while Claude coordinates a bounded revision loop. For automatic reviewer selection from either host, use claudex-loop with mode=review.
日本語の概要は準備中です。原文の説明を表示しています。
SUPERSEDED by /claudex-loop (formerly /crucible; adds a Phase 0 research/recon pass + rebuilt interview) — prefer that skill unless you explicitly want this one. Two-act plan hardening. ACT 1 (you ↔ Claude) — Claude interviews you relentlessly about a plan or design, one question at a time, recommending an answer for each and exploring the codebase when it can answer itself, until every branch of the decision tree is resolved. ACT 2 (Claude ↔ Codex) — Claude writes the locked plan to PLAN.md and OpenAI Codex adversarially reviews it in a read-only sandbox (VERDICT:APPROVED/REVISE), Claude revises and re-submits to the SAME Codex session until APPROVED or a MAX_ROUNDS cap, then you sign off before any code. Use when the user says "/grill-me-codex", "grill me then have codex review", "grill me and stress-test the plan", "interview me about this plan then get a second model on it", or is about to build something high-stakes (auth, schema, concurrency, migrations, payments) and wants both alignment AND a cross-model sanity check before implementation. Builds on Matt Pocock's grill-me (MIT). For the docs-aware variant use /grill-with-docs-codex; if you already have a plan and want only the Codex review use /codex-review. NOT for reviewing already-written code (use /codex:review) and NOT for trivial changes.
日本語の概要は準備中です。原文の説明を表示しています。
SUPERSEDED by /claudex-loop (formerly /crucible; its docs-aware mode covers this variant) — prefer that skill unless you explicitly want this one. Two-act plan hardening with living documentation. ACT 1 (you ↔ Claude) — Claude interviews you relentlessly about a plan, one question at a time, challenging it against your project's existing domain model and glossary (CONTEXT.md), sharpening fuzzy terms, stress-testing with concrete scenarios, cross-referencing code, and updating CONTEXT.md + ADRs inline as decisions crystallise. ACT 2 (Claude ↔ Codex) — Claude writes the locked plan to PLAN.md and OpenAI Codex adversarially reviews it in a read-only sandbox (VERDICT:APPROVED/REVISE), Claude revises and re-submits to the SAME Codex session until APPROVED or a MAX_ROUNDS cap, then you sign off before any code. Use when the user says "/grill-with-docs-codex", "grill me against the docs then have codex review", "stress-test this against our domain model then get a second model on it", or is about to build something high-stakes in a project with established terminology/ADRs and wants alignment, documentation, AND a cross-model sanity check. Builds on Matt Pocock's grill-with-docs (MIT). NOT for reviewing already-written code (use /codex:review) and NOT for trivial changes.
日本語の概要は準備中です。原文の説明を表示しています。