shawn-sandy/commit-agent
kit/plugins/git-agent/skills/commit-agent/SKILL.md
Stages all changes and creates a conventional commit message. Analyzes the diff and writes a descriptive, scope-correct commit. Use when the user asks to commit or save work to git.
$ install --global
skillsauthnpx skillsauth add shawn-sandy/agentics commit-agentInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 4, 2026, 2:38 AM40.3s1 file scanned
SKILL.md
- name:
- commit-agent
- description:
- Stages all changes and creates a conventional commit message. Analyzes the diff and writes a descriptive, scope-correct commit. Use when the user asks to commit or save work to git.
- allowed-tools:
- Bash(git *), ToolSearch, ExitPlanMode
- disable-model-invocation:
- true
Stage all changes and create a conventional commit message. Follow these steps in strict order. STOP immediately after step 4.
When not to use
Does not push or create PRs — use pr-agent for that.
Step 0: Exit Plan Mode
Call ExitPlanMode immediately and silently — always, unconditionally, before
any other action. Do not prompt the user. Staging and committing are
git mutations and cannot proceed inside plan mode.
ExitPlanMode is a deferred tool whose schema must be loaded before it can be
called. Use ToolSearch with select:ExitPlanMode first, then call
ExitPlanMode. Both steps happen silently with no user-visible output.
Error handling: If ExitPlanMode returns the exact error "You are not in plan mode", treat that as success — plan mode was already off. Do not abort or surface the error to the user; continue to the next step.
Step 1: Guards
Run git status to check repository state.
- Clean working tree (nothing to commit): output "Nothing to commit — working tree is clean." and STOP.
- Detached HEAD (
git branch --show-currentreturns empty): output "Cannot commit: repository is in detached HEAD state. Checkout a branch first." and STOP.
Step 2: Stage Changes
Run git add -A to stage all changes.
This trusts .gitignore to exclude sensitive or generated files. The user is responsible for .gitignore correctness.
Step 3: Analyze Diff and Write Commit Message
Run git diff --staged to inspect all staged changes.
Write a conventional commit message:
<type>(<scope>): <description>
Rules:
- Total length: ≤ 72 characters
- Types:
feat,fix,docs,refactor,test,chore,perf,style,ci,build - Scope: the most-changed top-level directory (e.g.,
plugins/git-agent→plugins/git-agent) - Omit scope entirely if changes span more than 2 top-level directories
- Description: imperative mood, lowercase, no trailing period
Examples:
feat(plugins/git-agent): add commit-agent and pr-agent skillsfix(plugins/code-review): correct activation trigger wordingchore: update marketplace.json with new plugin entry
Step 4: Commit
Run:
git commit -m "<message>"
Output the commit hash and message on success.
If a pre-commit hook fails: report the hook's output verbatim and STOP. Do not retry. Do not use --no-verify. Do not modify the staged files. Let the user fix the issue.
After a successful commit, output one line:
To undo:
git reset HEAD~1
STOP here. Do not run tests, analyze coverage, check for issues, push, create PRs, or take any further action.
Related Skills
shawn-sandy/craft-prompt
data-ai
VerifiedTrustedCommunity
Craft-prompt: interviews users and assembles a structured AI prompt using Anthropic best-practice techniques. Use when the user runs /plan-agent:craft-prompt or asks to craft a prompt.
2SKILL.mdUpdated Jun 5, 2026
shawn-sandy/share-init
development
VerifiedTrustedCommunity
Generates a SOCIAL.md project sharing config by analyzing the codebase. Use when asked to set up social sharing preferences or create a SOCIAL.md file.
2SKILL.mdUpdated Jun 4, 2026
shawn-sandy/share-explanation
development
VerifiedTrustedCommunity
Explains how any project file, component, or concept works. Reads source files and synthesizes developer-friendly principles, social copy, and a dark-mode card. Use when asked 'how does X work' or 'explain X'.
2SKILL.mdUpdated Jun 4, 2026
shawn-sandy/implementation-plan
development
VerifiedTrustedCommunity
Generate an HTML implementation-plan document. Produces a self-contained .html plan file with steps, acceptance criteria, and metadata. Use when the user asks to create a plan document, generate an HTML plan, or write a plan file — not for general planning questions.
2SKILL.mdUpdated Jun 4, 2026