The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using git and gh to manage pull request workflows.
Fetches repository data using git fetch origin <base> and generates diffs using git diff in references/shared.md.
Interacts with GitHub to retrieve pull request metadata using gh pr view and updates descriptions using gh pr edit as described in references/shared.md and references/describe.md.
Shell commands incorporate user-supplied variables (e.g., <base> branch names) or values retrieved from CLI outputs.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks common to tools that process untrusted external data like code diffs.
Ingestion points: The skill ingests untrusted content from git diff output, which may contain malicious instructions embedded in code comments or commit messages (identified in references/shared.md).
Boundary markers: There are no explicit instructions or delimiters used to warn the agent to ignore or isolate instructions found within the diff content.
Capability inventory: The skill has the ability to write to the filesystem and execute network-connected shell commands via gh.
Sanitization: No evidence of filtering or sanitization of the diff content is present before it is analyzed for summary generation.

pull-request