The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes pull request review comments which are untrusted external inputs (potentially from third-party contributors). These comments are used to generate code modifications, creating a surface where a malicious comment could attempt to trick the agent into introducing vulnerabilities, backdoors, or performing unauthorized git operations.
Ingestion points: Fetches review comments from GitHub as part of its core logic.
Boundary markers: The skill explicitly requires developer approval (@agent fix #N) before any fix is applied, which serves as a critical defense-in-depth measure.
Capability inventory: The skill possesses the ability to read and edit files, and execute shell commands via git and gh.
Sanitization: No specific sanitization or filtering of the comment content is mentioned prior to LLM processing.
[COMMAND_EXECUTION]: The skill is configured to execute shell commands using the git and gh binaries to manage the PR workflow.
Evidence: The YAML frontmatter specifies Bash(git *) and Bash(gh *) under the permissions.allow section.

pr-review