Git Commit Workflow

Use when the user wants a single clean commit created from current changes with safe staging, message drafting, and non-interactive git usage.

Security Scan

The skill's instructions and requirements are coherent for creating a single safe git commit; it is instruction-only, requests no credentials, and does not perform unexpected network or filesystem actions — but it assumes the agent can inspect and modify a git repository (the presence of git is not declared and provenance cannot be fully verified).

Purpose & Capability

The skill's name and description match its instructions: staging, drafting a message, and creating a non-interactive commit. Minor mismatch: the skill does not declare that the 'git' binary is required, yet the workflow implicitly needs git and write access to the repository.

Instruction Scope

SKILL.md stays within scope: it instructs the agent to inspect repo state (diff, status, branch), stage relevant changes, craft a message, and commit. It explicitly forbids amending, bypassing hooks, and committing secrets. There are no instructions to read unrelated system files or send data externally.

Install Mechanism

No install spec and no code files are present (instruction-only). This minimizes installation risk — nothing will be downloaded or written by the skill itself.

Credentials

The skill requires no environment variables, credentials, or config paths. Its needs (repository read/write access and a git binary at runtime) are proportionate to the stated purpose.

Persistence & Privilege

The skill does not request always:true or other elevated persistence. It is user-invocable and can be invoked autonomously per platform default, which is expected for a commit workflow skill.

Assessment

This skill appears to do what it says: stage and create a single non-amended commit. Before installing or running it, ensure the agent environment has git available and that you are comfortable allowing the agent to modify the repository (it needs write access). Because the package is instruction-only and the provenance points to a source file name but there's no code to inspect, you should: (1) verify the agent's actual runtime commands when first invoked (have it show the git commands it intends to run), (2) review the staged changes before confirming the commit to avoid accidental inclusion of secrets or unrelated files, and (3) require explicit permission for any amend/hook-bypass actions. If you need stronger assurance, request a skill version with verifiable source code or a declared dependency on git in its metadata.

Use this skill when the task is to create one commit from current work.

Workflow

1. Inspect git status, diff, branch, and recent commit style.
2. Stage only relevant changes.
3. Draft a concise commit message that reflects why the change exists.
4. Create a new commit non-interactively.
5. Report the resulting commit summary.

Guardrails

• Never amend unless explicitly asked.
• Never bypass hooks unless explicitly asked.
• Avoid staging secrets or unrelated files.
• Do not create an empty commit.

Example Requests

• Make one clean commit from the current working tree.
• Stage the right files and create a proper commit message.

Inputs

• Current diff
• Branch state
• Recent commit style

Outputs

• One non-amended commit
• Accurate commit message

Success Criteria

• Only relevant changes were committed.
• No empty commit was created.
• Hooks and safety rules were respected.

Non-Goals

• Amend flows unless explicitly asked
• Committing secrets or unrelated changes

Source Provenance

Derived from src/commands/commit.ts.