The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from GitHub Check outputs, which could contain malicious instructions designed to influence the agent.
Ingestion points: In fetch_failing_checks.py, the skill retrieves check run details, including title and summary fields, via the GitHub API.
Boundary markers: The diagnosis report generated in write_ci_fails_report.py incorporates this external data without using specific delimiters or instructions to treat the content as untrusted.
Capability inventory: The skill is highly autonomous, with the ability to automatically apply code fixes, commit changes, push to remote branches, and execute local verification scripts.
Sanitization: No sanitization or validation of the fetched check data is performed before it is added to the agent's context.\n- [COMMAND_EXECUTION]: The skill utilizes subprocess.run to execute shell commands, as seen in the _skill_cli.py helper. This capability is used to interact with the GitHub CLI and run discovered test scripts.\n- [DYNAMIC_EXECUTION]: The discover_and_verify.py script identifies and executes commands found in project configuration files like package.json, pyproject.toml, and Makefile. This represents dynamic execution where the exact commands run depend on the content of the target repository.

heyeddi-ci-fails