The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes Jenkins console output, which is an external data source that could contain malicious instructions.
Ingestion points: retrigger-ci.py fetches build logs from Jenkins via the fetch_console_tail function.
Boundary markers: The instructions do not define clear boundaries or provide explicit warnings to the agent when presenting parsed log data.
Capability inventory: The skill has the ability to execute shell commands (gh, git) and perform authenticated network operations (triggering Jenkins builds).
Sanitization: No sanitization is performed on the log content before it is parsed for test failure markers or displayed to the agent.
[COMMAND_EXECUTION]: The script executes local command-line tools to interact with GitHub and the repository.
Evidence: retrigger-ci.py uses subprocess.run to call the gh CLI, git grep, and a local analysis script check-upstream-flake.py.
Mitigation: The script uses list-based arguments instead of shell strings and validates input parameters, such as casting the PR number to an integer, which are established safe practices for process execution.

make-ci-green