The Agent Skills Directory

[COMMAND_EXECUTION]: The discover_and_verify tool executes standard build and test commands (such as npm test, pytest, cargo test, or make) found within the project repository. This execution is scoped to commands evidenced by existing project configuration files and is used to verify fixes before responding to PR findings.
[PROMPT_INJECTION]: The skill processes untrusted PR comments which could contain malicious instructions. To mitigate this risk, it implements a technical control in _untrusted_doc.py that wraps all external content in <<<UNTRUSTED_EXTERNAL_CONTENT>>> tags, explicitly instructing the agent to treat the text as data rather than instructions.
[COMMAND_EXECUTION]: A dedicated safety gate, assert_no_merge, monitors for merge-related commands and blocks execution unless a specific authorization phrase ("authorize merge") is provided by the user in the current session. This prevents unintended code merging during automated CI response cycles.
[EXTERNAL_DOWNLOADS]: The skill uses the gh (GitHub CLI) for all network interactions with GitHub repositories. These operations are standard authenticated API calls performed via established tooling.

heyeddi-ci-respond