The Agent Skills Directory

[COMMAND_EXECUTION]: The skill orchestrates multiple shell scripts and Git commands to perform repository health and security checks.
Orchestrates execution through scripts/preflight.sh which calls git fetch, git status, and git push.
Invokes scripts/scan-secrets.sh to perform automated security analysis of the pending commit range.
[EXTERNAL_DOWNLOADS]: The skill requires established third-party security utilities to function correctly.
References the official GitHub repository for gitleaks (github.com/gitleaks/gitleaks) for secret detection.
References the official GitHub repository for ripgrep (github.com/BurntSushi/ripgrep) for high-performance pattern matching.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Git diffs, creating a potential surface for indirect injection if a detected secret contains malicious instructions.
Ingestion points: scripts/scan-secrets.sh reads and processes output from git diff to identify secret patterns.
Boundary markers: Security findings are presented to the agent within a clearly labeled report section titled "SECRET-PATTERN MATCH (regex layer)".
Capability inventory: The skill utilizes Read, Bash, Glob, and Grep tools, and executes standard Git workflow commands.
Sanitization: The script includes basic output truncation (head -40) when reporting matches to the agent.

push-preflight