The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various CLI tools to manage the development lifecycle. It uses git for version control (staging, committing, and pushing), gh (GitHub CLI) to automate pull request creation, and nr (npm run) to verify code quality through testing, linting, and type-checking.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through its ingestion of external repository data.
Ingestion points: The skill reads and evaluates the content of AGENTS.md and branch diffs.
Boundary markers: There are no explicit markers used to separate untrusted data from instructions during the review phase.
Capability inventory: The agent has capabilities to modify the local filesystem, execute shell commands, and interact with remote repositories via git push and gh pr create.
Sanitization: The skill includes safety instructions that explicitly warn against committing secrets and require the agent to pause for user confirmation before applying changes that broaden scope or alter intended behavior.
[SAFE]: The skill's operations are consistent with its purpose as a development automation tool. Its reliance on human-in-the-loop checkpoints for critical decisions and its explicit instructions to avoid committing secrets mitigate the inherent risks of automated code processing.

ship