メニュー
Use when the user asks to run Claude Code CLI (`claude`, `claude resume`) for review, analysis, implementation, refactoring, debugging, or follow-up specifically through Claude Code.
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
| name | claude-code |
| description | Use when the user asks to run Claude Code CLI (`claude`, `claude resume`) for review, analysis, implementation, refactoring, debugging, or follow-up specifically through Claude Code. |
Use this skill to run Claude Code directly, capture Claude's actual output, and summarize it back to the user.
Mirror the codex skill's style: short, direct, and command-first. Do not improvise. Do not use the wrapper as the normal path.
| Model | Best for |
|---|---|
sonnet | Default recommendation for most coding, review, and analysis tasks |
opus | Stronger reasoning for ambiguous, high-stakes, or architecture-heavy work |
haiku | Fast, lightweight follow-ups and narrow questions |
default | Anthropic's account/environment default |
opusplan | Planning-heavy tasks that intentionally mix planning and execution |
Default recommendation: sonnet for most tasks, opus when depth matters more than speed.
Choose the permission mode yourself based on the task. Do not ask the user unless they explicitly want to control it.
| Mode | Use when |
|---|---|
default | Normal review, analysis, debugging, and repo inspection |
acceptEdits | Claude should edit files or implement changes |
plan | The user explicitly wants strict read-only planning / analysis |
bypassPermissions | Only with explicit user approval in a trusted environment |
Hard rule: do not use plan for ordinary reviews just because it sounds safe. For normal reviews, use default.
sonnet) in one short question, unless they already specified it.defaultacceptEditsplan--max-turns as an optional circuit breaker, not a default quality knob.
--max-turns3610 to 12--effort high by default. Use the CLI default unless the user explicitly wants deeper reasoning or the task is unusually ambiguous.claude command with the appropriate options:
-p, --print--output-format text for short runs that should complete quickly--output-format stream-json --verbose for long repo reviews, investigations, or any run where you need progress visibility--model <MODEL>--permission-mode <MODE>--max-turns <N> only when you intentionally want to cap the run"your prompt here" immediately after -p when using variadic flags like --allowedTools--allowedTools <TOOL> [<TOOL> ...] only to pre-approve tools without prompting--disallowedTools <TOOL> [<TOOL> ...] when you actually need to prevent specific toolsWhen this skill is used from Codex, use the execution path that actually works here.
exec_command with sandbox_permissions: "require_escalated".claude -p has been verified to work outside the sandbox and to hang or misbehave inside it.--output-format stream-json --verbose so progress is visible. Plain text output can remain silent for a long time even when Claude is actively working.--max-turns unless the user explicitly wants a hard cap. Claude may spend several turns exploring the repo before producing the review.Use this as the default review path:
claude -p \
"Review this repository and return findings ordered by severity with file references where possible." \
--output-format stream-json \
--verbose \
--model sonnet \
--permission-mode default \
--allowedTools "Read" "Grep" "Glob" "LS" "Bash(git status:*)" "Bash(git diff:*)" \
--disallowedTools "Edit" "Write" "NotebookEdit"
git diff --staged | claude -p \
"Review this diff and return findings ordered by severity with concise explanations." \
--output-format text \
--model sonnet \
--permission-mode default
Use this when the prompt names several directories or asks for deeper inspection:
claude -p \
"Inspect the relevant parts of this repo, evaluate the design, and return evidence-based findings ordered by severity." \
--output-format stream-json \
--verbose \
--model opus \
--permission-mode default \
--allowedTools "Read" "Grep" "Glob" "LS" "Bash(git status:*)" "Bash(git diff:*)" \
--disallowedTools "Edit" "Write" "NotebookEdit"
claude -p \
"Implement the requested change in this repo, keep the diff minimal, and summarize what changed." \
--output-format text \
--model sonnet \
--permission-mode acceptEdits
claude -r latest -p "Focus only on the open issue you identified earlier and propose the safest next step." \
--output-format text
claude --version or a claude -p command exits non-zero.max turns, report that plainly. Do not paraphrase it as a hang.--max-turns by default. If the user did not ask for a hard cap, leave it out.--output-format text stays silent during a long repo review, do not assume it is hung. Re-run or start with --output-format stream-json --verbose to confirm whether Claude is actively working.plan after a failed default run.--allowedTools, because this CLI can consume it as another allowed tool and then fail with "Input must be provided..."--allowedTools as a strict read-only sandbox. It pre-approves tools; it does not, by itself, ban other tools.cat file1 file2 ... | claude ... pipelines.Claude Code is a colleague, not an authority.
Survey a repo and ground externally on the relevant parts of the internet. Local manifest scan + CLAUDE.md/AGENTS.md + recent git activity, then external grounding via WebSearch, WebFetch, and gh — synthesized into a terse, dated, sourced briefing cached at docs/ground/<fingerprint>.md with a 7-day freshness window. Composable: invoked standalone or embedded as a pre-flight phase in other skills. Triggers: ground, ground yourself, ground myself, survey and ground, pre-flight grounding, research the stack, what's new in.
Collaborative discovery before planning. Explore the problem space, evaluate approaches, surface past work, and produce a structured brainstorm document. Triggers: brainstorm, explore, discovery, ideate, think through, what should we build, explore approaches.
Strategic planning with auto-calibrated detail, decision rationale, and dependency ordering. Use when starting a new feature, bug fix, refactor, or any non-trivial work. Produces a plan document with tasks, reasoning, and acceptance criteria. Triggers: plan, planning, create plan, implementation plan, feature plan, work plan.
Execute a plan from start to ship — read tasks, implement in dependency order, test continuously, commit incrementally, run quality checks, and push. The plan's checkboxes are the tracker. Triggers: work, execute plan, implement, start work, build, ship, finalize, release, push, ready to ship, done building.
Capture a footgun the agent just hit (or the user just spotted) as a proposed trigger via `quellis teach`. Guided flow: gather the finding, run the CLI, surface the proposed trigger, leave activation to the user. Triggers: teach quellis, capture this footgun, add a trigger, remember this rule.
Review and manage evidence contracts in this repo. Subcommands: `show` (default, print the active contract), `list` (all contracts + which is active), `complete` (mark active contract completed), `abandon` (mark active contract abandoned). Triggers: contract, show contract, list contracts, mark contract complete, what's active.