The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to automatically fetch and process content from external github.com links (issues, pull requests, and commits). This content originates from untrusted third parties and could contain instructions designed to influence the agent's behavior.
Ingestion points: Untrusted data enters the agent context via gh and git commands when inspecting GitHub URLs (found in SKILL.md instructions).
Boundary markers: The instructions do not specify any delimiters or safety warnings to ignore instructions embedded within the fetched GitHub content.
Capability inventory: The skill possesses extensive system capabilities including repository configuration, git history modification (rebase/bisect), and pull request management.
Sanitization: There is no evidence of sanitization or filtering applied to the data retrieved from external URLs.
[COMMAND_EXECUTION]: The skill operates primarily by executing shell commands via the git and gh binaries, which is the intended functionality but involves significant system interaction.
[DYNAMIC_EXECUTION]: The skill includes instructions to fix and maintain git hooks (found in SKILL.md under "Advanced git operations"). Git hooks are executable scripts that run locally; modifying them allows for the configuration of local code execution.

octocat