The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data originating from external GitHub pull requests which could potentially contain malicious instructions.
Ingestion points: Pull request titles, file metadata, and patch content retrieved using gh pr list and gh api in SKILL.md.
Boundary markers: Absent. The skill lacks specific instructions to delimit or ignore potentially malicious content embedded within the pull request data.
Capability inventory: The skill possesses the ability to execute shell commands, perform Git operations, use the GitHub CLI, and run application verification scripts (e.g., hitting health endpoints).
Sanitization: The skill applies literal string replacements for patches but does not explicitly validate or sanitize the PR metadata before using it in descriptive contexts.
[COMMAND_EXECUTION]: The skill relies on executing system-level commands to interact with version control and the GitHub API.
Uses git for branch management, fetching, and applying code changes.
Uses gh (GitHub CLI) for PR lifecycle management and repository metadata retrieval.
Instructs the agent to run verification steps including linting, type-checking, and booting the application to test health endpoints.

dependabot-backlog