The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions for verifying pull requests require the agent to "check it out, run the relevant tests or commands." This process involves executing code provided by external, untrusted contributors, which could allow a malicious actor to run arbitrary commands in the agent's environment.
[PROMPT_INJECTION]: The skill is designed to ingest and process data from external issue trackers, such as issue bodies, comments, and pull request diffs. This creates a surface for indirect prompt injection attacks where malicious instructions embedded in an issue could override the agent's logic or influence its triage decisions.
Ingestion points: Full issue and pull request bodies, comments, and diff content parsed during context gathering (SKILL.md).
Boundary markers: None identified. There are no instructions for the agent to use delimiters or to disregard instructions found within the data being triaged.
Capability inventory: Reading files across the codebase, checking out pull requests, executing arbitrary commands and tests, posting comments to the issue tracker, and modifying repository files like .out-of-scope/*.md and CONTEXT.md (SKILL.md, OUT-OF-SCOPE.md).
Sanitization: None identified. The agent is instructed to "Interpret the request and act" based on natural language descriptions provided in the tracker.

triage