The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data from issue trackers and pull requests to automate triage tasks.
Ingestion points: The agent reads the body, comments, and diffs of issues and pull requests from external sources in SKILL.md (Step 1: Gather context).
Boundary markers: The instructions lack specific delimiters or instructions to ignore or isolate embedded prompts within the external data.
Capability inventory: The agent has the ability to search the codebase, execute tests and commands to verify PRs (SKILL.md Step 3), post comments to the issue tracker, and modify repository files such as GLOSSARY.md and the .out-of-scope/ directory.
Sanitization: No sanitization or filtering logic is defined to protect the agent from malicious instructions hidden in issue descriptions or PR comments.
[COMMAND_EXECUTION]: The skill instructs the agent to execute code provided by external, untrusted sources as part of its verification workflow.
Evidence: In SKILL.md (Step 3: Verify the claim), the instructions direct the agent to "check it [the PR] out, run the relevant tests or commands" to confirm the diff's behavior. This presents a risk of executing malicious commands if a contributor embeds harmful code within the pull request's tests or verification steps.

triage