The Agent Skills Directory

[SAFE]: No malicious behavior detected. The skill automates standard PR preparation tasks like running tests and linting.
[COMMAND_EXECUTION]: The skill invokes local project commands such as make, npm, and pytest to execute quality gates before PR creation.
[EXTERNAL_DOWNLOADS]: Utilizes modular dependencies (e.g., sanctum:git-workspace-review, scribe:slop-detector) for shared functionality and content validation.
[DATA_EXFILTRATION]: The skill records execution details, including commands and file paths, to a local audit log file (/tmp/skill-audit.log) for tracking progress.
[PROMPT_INJECTION]: The skill processes potentially untrusted data from git history and diffs.
Ingestion points: Git diff output and file contents are read to generate summaries in SKILL.md (Step 3).
Boundary markers: The instructions do not define delimiters to isolate diff content from task instructions.
Capability inventory: Access to Bash for command execution and Write for file modifications.
Sanitization: Ingested git data is processed without explicit sanitization or filtering.

pr-prep