The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes gh (GitHub CLI) commands to fetch repository data. This is an expected behavior for a synchronization tool and uses an official, well-known utility from a trusted organization.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted data from external sources.
Ingestion points: Data is pulled from GitHub PR titles, descriptions (bodies), and branch names via gh pr list and gh pr view.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the processed PR content are defined.
Capability inventory: The agent can execute local gh CLI commands and has write access to the pm-workspace-docs/ directory to update metadata and reports.
Sanitization: There is no evidence of sanitization or escaping of the PR body text before it is interpolated into summaries or used for metadata updates.

github-sync