claude-review
Launch an isolated Claude reviewer for the current branch, a specific commit, or uncommitted changes. Uses Claude's agent runtime with worktree isolation so the review has an independent read-only view of the code before findings are reported back.
What this skill does
<EXTREMELY-IMPORTANT>
This skill is review orchestration, not code editing.
Non-negotiable rules:
1. Read the real diff before building the reviewer prompt.
2. Always use isolated review execution.
3. Tell the reviewer to report only, never modify code.
4. Carry forward exclusion lists on repeated rounds so fixed findings are not re-reported.
5. Verify returned findings before acting on them.
</EXTREMELY-IMPORTANT>
# Claude Review
## Inputs
- `$request`: Optional scope hint such as `last commit`, `uncommitted`, `auth focus`, or `round 2`
## Goal
Create a focused Claude review that:
- derives its scope from the real diff
- gives the reviewer the right focus areas
- runs in isolation from the main working tree
- returns actionable findings without changing code
## Step 0: Resolve the review scope
Determine whether the user wants:
- full branch review
- last-commit review
- uncommitted review
- targeted review of a subsystem or recent task
Read the relevant diff summary and changed-file list before building the prompt.
If there is nothing to review, stop and say so explicitly.
**Success criteria**: The review scope is explicit and backed by a real diff.
## Step 1: Build a focused review brief
Use the diff to identify:
- changed files
- changed directories or subsystems
- likely risk areas such as auth, secrets, concurrency, shelling out, migrations, or tests
- issues already fixed in previous rounds that should not be re-reported
The prompt should include:
- review base and scope
- changed-file list
- focused risk areas
- explicit report-only instruction
- desired output format
**Success criteria**: The reviewer brief is specific to the actual changes, not generic boilerplate.
## Step 2: Launch the isolated reviewer
Use `Agent` with:
- `subagent_type: general-purpose`
- `isolation: "worktree"`
- a prompt that says:
- read the full diff
- read surrounding file context
- verify each finding before reporting
- do not make code changes
- report findings as a markdown table:
`| # | Priority | File:Line | Description | Suggested Fix |`
- priority levels: P1 (crash/security), P2 (incorrect behavior), P3 (minor)
- if nothing significant is found, say so explicitly
If the review is expected to take a while, background execution is acceptable, but keep the scope tight enough that the reviewer stays focused.
**Success criteria**: The review runs in an isolated worktree and cannot mutate the active working copy.
## Step 3: Parse and summarize findings
Report:
- scope reviewed
- findings grouped by priority
- file and line references
- explicit clean result when no material findings are returned
Treat reviewer output as candidate findings, not automatic truth. If the user wants fixes, verify each finding locally first.
**Success criteria**: The user gets a readable review summary without needing to parse raw agent output.
## Step 4: Iterate only when useful
On a second or later round:
- include previously fixed findings in the exclusion list
- tighten the review focus to new or modified areas
- avoid rerunning a generic full-branch review if only a few files changed
**Success criteria**: Each new round looks for genuinely new issues instead of rehashing old ones.
## Comparison with Sibling Skills
| Skill | Tool | Best For |
|-------|------|----------|
| `/claude-review` | Claude Code (Agent) | Deep code understanding, context-aware review |
| `/codex-review` | OpenAI Codex CLI | Independent perspective, repro scripts |
| `/kiro-review` | Kiro CLI | Alternative AI perspective |
For critical code, run all three and cross-reference findings. The overlap between reviewers catches more bugs than any single tool.
## Guardrails
- Do not add `context: fork`; this skill already uses the agent runtime directly.
- Do not let the reviewer edit code.
- Do not skip diff reading before prompt construction.
- Do not include real secret values in the prompt or summary.
- Do not run this skill proactively; it is explicit-user-only.
## Output Contract
Report:
1. review scope
2. changed areas emphasized in the prompt
3. findings by priority with `file:line`
4. explicit clean result if no significant findings were returned
5. whether a later review round should exclude prior fixed issues
Related in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.