The Agent Skills Directory

[COMMAND_EXECUTION]: Orchestrates pull request workflows using vendor-internal tools (push-pr, wait-ci, and fix-pr).
[PROMPT_INJECTION]: The skill implements an automated feedback loop that ingests untrusted data from CI failures and pull request comments to drive code modifications.
Ingestion points: CI status and review comments retrieved by codagent:wait-ci in SKILL.md.
Boundary markers: No explicit boundary markers or isolation instructions are defined in the orchestration logic.
Capability inventory: The codagent:fix-pr tool has the capability to modify the codebase based on the ingested input.
Sanitization: No sanitization or validation of external comments is mentioned before they are processed by the fixing tool.
[SAFE]: Employs explicit safety termination rules, including a maximum limit of three fix cycles and detection of persistent failure signatures, to prevent runaway resource usage or repetitive errors.

finalize-pr