The Agent Skills Directory

[COMMAND_EXECUTION]: The skill script scripts/p9.py interacts with development tools and system utilities via the subprocess module to manage pull request lifecycles.\n
Evidence: The script invokes gh (GitHub CLI), git, bun, and pytest to perform tasks such as watching PR checks, merging branches, and executing build or test cycles.\n- [COMMAND_EXECUTION]: The heal subcommand executes remediation commands using subprocess.run with shell=True.\n
Evidence: These commands (e.g., bun run lint:fix) are sourced from a pre-defined rubric in the code and are only triggered when failure logs match specific regex patterns. This execution is limited to a set of hardcoded development tasks.\n- [COMMAND_EXECUTION]: The conformance command allows the test runner to be customized through the BROOMVA_P9_PYTEST environment variable.\n
Evidence: The environment variable is split into a command list and executed, providing a standard mechanism for developers to specify their preferred test environment.\n- [SAFE]: The skill processes external data in the form of CI build logs to determine failure categories and remediation actions.\n
Ingestion points: Output from gh run view --log-failed or input provided via standard input.\n
Boundary markers: None.\n
Capability inventory: Execution of pre-defined development commands via subprocess.\n
Sanitization: Remediation commands are selected from a static list in the code based on regex matching and are not constructed from the content of the builds logs, preventing command injection via build output.

p9