The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from the local repository (git diffs, commit subjects, and file contents) to generate PR summaries and descriptions. \n
Ingestion points: Git diff output, git log entries, and repository files analyzed in Step 3 and Step 2.5 of SKILL.md. \n
Boundary markers: No explicit delimiters or instructions are used to differentiate untrusted repository data from the agent's instructions when generating the PR description. \n
Capability inventory: The skill can execute shell commands (grep, make, npm, etc.), run Python scripts, and write files to the local file system. \n
Sanitization: Repository content is not explicitly sanitized before being used to populate the PR template, allowing for malicious instructions in the code or history to potentially influence the agent's behavior. \n- [COMMAND_EXECUTION]: The skill directs the agent to execute project-specific commands for quality assurance, such as make test, npm run lint, and cargo clippy. While standard for development, this relies on the safety of the repository's configuration. \n- [DYNAMIC_EXECUTION]: The workflow involves running a local analysis script (pr_prep_analyze.py) located in the plugin's root directory to process changes and recommend merge strategies.

pr-prep