The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted issue and pull request data from GitHub, which could contain malicious instructions designed to influence the agent's behavior during the design phase. * Ingestion points: The skill ingests data via issues.json (exported from GitHub) and through instructions to the agent to manually sample open issues. * Boundary markers: The skill does not provide specific delimiters or 'ignore' instructions for the sampled data, leaving it open to interpretation by the agent. * Capability inventory: The skill uses shell command execution (gh, python3) and file reading to generate reports and propose policy changes. * Sanitization: The triage-baseline.py script uses standard json and csv libraries for safe parsing, but there is no sanitization or isolation layer for content read manually by the agent.
[COMMAND_EXECUTION]: The skill utilizes shell commands (gh issue list and python3) to export and process repository data. These are standard maintenance operations but represent the execution of commands on the host system based on the skill's instructions.

oss-issue-triage