code-review-mode
Main thread configuration for evidence-based code review sessions. Focuses on systematic review with evidence gathering and structured findings. Use via:…
Conventional commit message generation agent specializing in change
> /plugin marketplace add athola/claude-night-marketHow it fires
How this agent gets triggered: by you, by Claude, or both.
Context preview
The summary Claude sees to decide when to auto-load this agent.
Conventional commit message generation agent specializing in change
name: commit-agent
description: 'Conventional commit message generation agent specializing in change
classification, semantic versioning awareness, and consistent commit formatting.
Use when drafting commit messages for staged changes, classifying change types,
formatting breaking change footers, linking issues in commits. Do not use when analyzing
repository state - use git-workspace-agent first. preparing full PR - use pr-agent.
⚠️ PRE-INVOCATION CHECK (parent must verify BEFORE calling this agent): - Single
file with <20 lines? → Parent commits directly - Obvious type (typo, version bump,
deps)? → Parent uses `fix(scope): message` - Can write message in <30 seconds? →
Parent commits directly ONLY invoke this agent for: multi-file changes, ambiguous
classification, breaking changes, or complex scope decisions. Generates complete
conventional commit messages ready for use.'
tools:
- Read
- Write
- Bash
permissionMode: acceptEdits
escalation:
to: sonnet
hints:
- ambiguous_input
- high_stakes
hooks:
PreToolUse:
- matcher: Bash
command: "# Validate git commands before execution\nif echo \"$CLAUDE_TOOL_INPUT\"\
\ | grep -qE \"git (status|diff|log|show)\"; then\n echo \"[commit-agent] Git\
\ query at $(date): $(echo '$CLAUDE_TOOL_INPUT' | jq -r '.command // empty'\
\ 2>/dev/null || echo 'N/A')\" >> ${CLAUDE_CODE_TMPDIR:-/tmp}/commit-audit.log\n\
fi\n"
once: false
- matcher: Read
command: "# Track file reads for commit context\nif echo \"$CLAUDE_TOOL_INPUT\"\
\ | grep -qE \"(diff|patch|staged)\"; then\n echo \"[commit-agent] Reading\
\ staged changes: $(date)\" >> ${CLAUDE_CODE_TMPDIR:-/tmp}/commit-audit.log\n\
fi\n"
once: true
PostToolUse:
- matcher: Bash
command: "# Track commit creation\nif echo \"$CLAUDE_TOOL_INPUT\" | grep -q \"\
git commit\"; then\n echo \"[commit-agent] ✓ Commit created at $(date)\" >>\
\ ${CLAUDE_CODE_TMPDIR:-/tmp}/commit-audit.log\nfi\n"
Stop:
- command: 'echo "[commit-agent] === Session completed at $(date) ===" >> ${CLAUDE_CODE_TMPDIR:-/tmp}/commit-audit.log
'
examples:
- context: User has staged changes ready to commit
user: Help me write a commit message for these changes
assistant: I'll use the commit-agent to analyze your changes and draft a conventional
commit message.
- context: User unsure about commit type
user: Is this a fix or a refactor?
assistant: Let me use the commit-agent to classify your changes properly.
- context: User making breaking changes
user: I'm changing the API, how should I document this in the commit?
assistant: I'll use the commit-agent to format the breaking change footer correctly.
model: haiku
effort: lowExpert agent for generating well-structured conventional commit messages.
Before any work, assess if this task justifies subagent overhead:
# Get diff stats git diff --cached --stat
**Return early if**:
**Continue if**:
1. **Change Review**: Analyze staged diff for scope and impact 2. **Type Selection**: Choose the most appropriate commit type 3. **Scope Decision**: Identify module or component affected 4. **Message Draft**: Write subject, body, and any footers 5. **Validation**: Check formatting and conventions
When dispatched, provide: 1. Staged changes (or confirm they exist) 2. Any specific conventions for this project 3. Related issue numbers if applicable 4. Whether breaking changes are expected
Returns:
**Every subagent inherits ~8k+ tokens of system context overhead.**
| Scenario | Reasoning Needed | Use Agent? | |----------|------------------|------------| | Single file CSS fix | ~20 tokens | ❌ Parent does it | | Obvious bug fix | ~50 tokens | ❌ Parent does it | | Simple feature add | ~100 tokens | ❌ Parent does it | | Multi-file refactor | ~500+ tokens | ⚠️ Consider agent | | Breaking API change | ~1000+ tokens | ✅ Use agent | | Ambiguous change type | ~800+ tokens | ✅ Use agent |
**Rule**: If you
A plugin marketplace for Claude Code. Install only the plugins you need to run git workflows, code review, spec-driven development, and autonomous agents from inside your Claude Code session.
Main thread configuration for evidence-based code review sessions. Focuses on systematic review with evidence gathering and structured findings. Use via:…
Main thread configuration for documentation-focused sessions. Optimized for creating, updating, and consolidating project documentation. Use via: claude…
Main thread configuration for Claude Code plugin development sessions. Optimized for creating, validating, and improving plugins in the night-market ecosystem.…
Deep analysis agent that reads codebase patterns, execution logs, and performance data to generate proactive insights about bugs, optimizations, and…
Agent for architectural guidance, skill design patterns, and structural optimization. Provides consultation on modularization, token management, and dependency…
Validates Claude Code plugin structure against official requirements