The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub issues, pull requests, and discussions. Ingestion points include yeet-context.sh (via GraphQL), gh discussion view, and gh pr view which fetch user-generated content from external repositories. While the skill lacks explicit boundary markers or 'ignore instructions' delimiters when interpolating this untrusted data into prompts for title/body generation, it defines an 'External-disclosure Review' protocol in references/posting.md, instructing the agent to review all generated content for sensitive data before posting. Capabilities include repository write access and git push functionality.
[EXTERNAL_DOWNLOADS]: The scripts/issue-form.py script uses the uv tool to manage dependencies, downloading the pyyaml package from PyPI. This is a standard dependency from a well-known registry.
[COMMAND_EXECUTION]: The skill uses gh and git CLI tools to interact with GitHub. It performs validation on repository names and template paths in yeet-context.sh and scripts/issue-form.py to reduce the risk of command injection.

yeet