The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard git and gh CLI tools to manage pull requests, which is consistent with its stated purpose. These operations include fetching branch status, viewing diffs, and pushing code to origin.
[COMMAND_EXECUTION]: Implements dynamic context injection using the !command`` syntax in SKILL.md to pre-load the agent's context with repository information. These commands (e.g., git status, git log) are diagnostic and do not directly incorporate unvalidated user input into the shell execution path at load time.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests untrusted data from the git repository.
Ingestion points: Data from git log and git diff is interpolated into the agent's context via dynamic commands in SKILL.md.
Boundary markers: While content is wrapped in tags like <full_diff>, there are no explicit instructions or delimiters telling the agent to ignore instructions embedded within the git data.
Capability inventory: The agent possesses the capability to perform write operations via git push and gh pr create/edit.
Sanitization: The skill does not perform any sanitization or filtering of commit messages or code diffs before they are processed by the AI.

pr