The Agent Skills Directory

[SAFE]: The skill is designed with a defensive posture, requiring explicit user authorization for all Git mutations and enforcing verification of staged diffs before commits.
[EXTERNAL_DOWNLOADS]: The documentation mentions installation and update procedures via npx using the author's own GitHub repository (github.com/idaibin/aicraft). This is a standard mechanism for skill distribution and follows vendor naming conventions.
[COMMAND_EXECUTION]: The skill facilitates the execution of local Git commands (git status, git fetch, git commit, git push, etc.) which are necessary for its intended purpose of repository management.
[PROMPT_INJECTION]: The skill processes external repository data which serves as a potential surface for indirect prompt injection.
Ingestion points: Reads AGENTS.md files and Git diffs from the local repository (SKILL.md, checklist.md).
Boundary markers: Explicitly instructs the agent to verify the staged diff and obtain direct user approval before performing mutations.
Capability inventory: Git mutations including staging, committing, and pushing.
Sanitization: Relies on human-in-the-loop verification and validation commands to ensure the integrity of the delivery process.

code-delivery