cross-review を独立 reviewer session として再定義する

AGENTS.md

@@ -65,23 +65,23 @@ These strings are not localizable in the current implementation. Forking is requ
 - Branch naming is the skill's responsibility (LLM-named in kebab-case, or user-supplied verbatim). The `worktree-` prefix forced by `EnterWorktree` is intentionally accepted; the `path` parameter escape hatch is out of scope (see issue #13).
 - The skill is a no-op when the current session is already inside a worktree — `EnterWorktree` itself rejects re-entry, and the skill double-checks via `git rev-parse --git-common-dir` / `--git-dir` before calling the tool.
 
-## Cross-review backend selection
+## Cross-review reviewer session selection
 
-`cross-review` supports two backends, selected via the `CROSS_REVIEW_BACKEND` environment variable:
+`cross-review` is defined as a second-opinion review from an independent reviewer session, not as a guarantee that a different backend or different model is used.
 
-- `codex` — OpenAI Codex CLI (`codex exec` with stdin diff pipe; the `review` sub-command is avoided because of the 0.125.0 `--base` / `--uncommitted` / `[PROMPT]` mutual exclusion). Intended for "implemented with Claude Code → reviewed by GPT".
-- `claude-self` — Claude CLI headless (`claude -p` with stdin diff). Intended for "implemented with Codex CLI / Cursor → reviewed by Claude".
+- Codex runtime uses Codex CLI (`codex exec --sandbox read-only` with stdin diff pipe) to start a fresh reviewer session.
+- Claude Code runtime uses Claude CLI headless (`claude -p` with stdin diff) to start a fresh reviewer session.
 
-When the env var is unset, the skill falls back to `command -v` auto-detection (`codex` first, then `claude`). When the env var is **set** but the corresponding CLI is missing, the skill fails explicitly — there is no silent fallback to the other backend, since that would silently change the reviewer model the user asked for.
+The runtime must be determined from the running agent's explicit environment, not inferred from whichever CLI exists on `PATH`. Environment-variable backend overrides and auto-detection fallback are intentionally not part of the workflow. If a different backend / different model review is needed, track that as a separate issue instead of keeping it inside `cross-review`.
 
 ## Cross-review base branch resolution
 
-`cross-review` resolves the base branch dynamically via `gh repo view --json defaultBranchRef --jq '.defaultBranchRef.name'` and feeds the result into `git diff "$BASE_REF"...HEAD` for both backends (the diff is piped to `codex exec` / `claude -p` via stdin). `master` / `develop` / `trunk` repos work without modification. The skill stops with an explicit error (no silent fallback to `main`) when default-branch resolution fails — see its "失敗時の対応" section. Override (env var / arg) is intentionally out of scope.
+`cross-review` resolves the base branch dynamically via `gh repo view --json defaultBranchRef --jq '.defaultBranchRef.name'` and feeds the result into `git diff "$BASE_REF"...HEAD` (the diff is piped to `codex exec` / `claude -p` via stdin). `master` / `develop` / `trunk` repos work without modification. The skill stops with an explicit error (no silent fallback to `main`) when default-branch resolution fails — see its "失敗時の対応" section. Override (env var / arg) is intentionally out of scope.
 
 ## External dependencies
 
 - `gh` CLI — all GitHub operations. Must be authenticated against the target repo.
-- At least one of: Codex CLI (`brew install --cask codex`) or Claude CLI (`npm install -g @anthropic-ai/claude-code`), required by `cross-review`. The skill must fail loudly (not silently skip) when neither is available, or when an explicitly-selected backend's CLI is missing.
+- The CLI for the current agent runtime: Codex CLI (`brew install --cask codex`) when implementing from Codex, or Claude CLI (`npm install -g @anthropic-ai/claude-code`) when implementing from Claude Code. `cross-review` must fail loudly (not silently skip) when the corresponding CLI is unavailable or the current runtime has no documented reviewer-session launch step.
 - Claude Code v2.1.49 or newer — required by `worktree-start` for the `EnterWorktree` tool. Older versions surface this as "tool not found"; the skill instructs users to upgrade rather than attempting any workaround.
 
 ## Editing skills

README.md

@@ -73,14 +73,14 @@ This installs the skills under your agent's skill directory (e.g. `~/.claude/ski
 
 issuekit assumes the following tools are available on the host:
 
-- **An [Agent Skills](https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview)-compatible agent runtime** (e.g. [Claude Code](https://docs.claude.com/en/docs/claude-code), Codex CLI, Cursor) that loads the skills.
+- **An [Agent Skills](https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview)-compatible agent runtime** (e.g. [Claude Code](https://docs.claude.com/en/docs/claude-code), Codex CLI, Cursor) that loads the skills. The full `issue-implement` cycle currently requires Codex CLI or Claude Code because `cross-review` has reviewer-session launch steps only for those runtimes.
 - **[`gh` CLI](https://cli.github.com/)** — used for all GitHub interactions (issue read/write, PR creation, CI status).
-- **At least one cross-review backend CLI** — required by `cross-review`. Pick whichever pairs with the agent driving the implementation:
-  - **[Codex CLI](https://github.com/openai/codex)** (`brew install --cask codex`) for the `codex` backend.
-  - **[Claude CLI](https://docs.claude.com/en/docs/claude-code)** (`npm install -g @anthropic-ai/claude-code`) for the `claude-self` backend (uses `claude -p` headless mode).
-- **Claude Code v2.1.49 or newer** — required by `worktree-start` only (it uses the `EnterWorktree` tool added in 2.1.49). The other six skills run on any Agent Skills-compatible runtime.
+- **The CLI for your current agent runtime** — required by `cross-review` to start an independent reviewer session:
+  - **[Codex CLI](https://github.com/openai/codex)** (`brew install --cask codex`) when the implementation is driven from Codex CLI.
+  - **[Claude CLI](https://docs.claude.com/en/docs/claude-code)** (`npm install -g @anthropic-ai/claude-code`) when the implementation is driven from Claude Code (uses `claude -p` headless mode).
+- **Claude Code v2.1.49 or newer** — required by `worktree-start` only (it uses the `EnterWorktree` tool added in 2.1.49). Other skills load on any Agent Skills-compatible runtime, but `cross-review` currently documents reviewer-session launch steps only for Codex CLI and Claude Code.
 
-`gh` must be authenticated against the repository you want to operate on. The `cross-review` backend is selected via the `CROSS_REVIEW_BACKEND` environment variable (`codex` / `claude-self`); if it is unset, the skill auto-detects whichever CLI is on `PATH`. If neither backend CLI is available, `cross-review` fails explicitly rather than silently skipping the review.
+`gh` must be authenticated against the repository you want to operate on. `cross-review` does not switch to another backend automatically; it uses the CLI that corresponds to the runtime currently driving the implementation. If that CLI is unavailable, or if the current runtime has no documented reviewer-session launch step, `cross-review` fails explicitly rather than silently skipping the review.
 
 ---
 
@@ -94,9 +94,9 @@ issuekit ships seven skills under `skills/`:
 | `issue-refine`       | Entry point | Re-shape an existing issue (title-only or partially formatted) into the standard format.                                                               |
 | `issue-pick`         | Entry point | Read-only triage: from a set of open issues, suggest the next one to take on, with rationale.                                                          |
 | `worktree-start`     | Entry point | **Claude Code only.** Switch the running session into a freshly named git worktree via the `EnterWorktree` tool. Accepts a task description **or** an issue URL/number — when the input is a `Status: Ready` issue, it chains into `issue-implement` after the worktree switch (otherwise it stops at the switch). |
-| `issue-implement`    | Orchestrator| Drive the full cycle from an issue number: status check → worktree start → implementation / commits → acceptance check → cross-review → PR → CI.     |
+| `issue-implement`    | Orchestrator| Drive the full cycle from an issue number: status check → worktree start → implementation / commits → acceptance check → cross-review → PR → CI. The full cycle currently requires Codex CLI or Claude Code because of `cross-review`. |
 | `acceptance-check`   | Verifier    | Read-only verifier that extracts `## 受け入れ条件` from an issue body and reports each item as `✓ / ✗ / ?`. Called by `issue-implement` after implementation/commits, before `cross-review`. |
-| `cross-review`       | Verifier    | Delegate a second-opinion code review to a different AI backend (Codex CLI or Claude CLI headless, selectable via `CROSS_REVIEW_BACKEND`) before PR creation. Called by `issue-implement` after `acceptance-check` passes; review fixes land as additional commits. |
+| `cross-review`       | Verifier    | Start an independent reviewer session with the current runtime's CLI and get a second-opinion code review before PR creation. Called by `issue-implement` after `acceptance-check` passes; review fixes land as additional commits. |
 
 `issue-implement` is the orchestrator; the other skills are either entry points or verifiers it calls. `worktree-start` is the only entry point that is Claude Code-specific (`EnterWorktree` is a Claude Code primitive — Codex CLI has no equivalent), so it has no fallback under other agent runtimes. It is also the only entry point that conditionally chains into the orchestrator: when invoked with an issue URL/number whose body has `Status: Ready`, it hands off to `issue-implement` after the worktree switch.