The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The connection strings identified in references/detection_rules.md (e.g., postgres://[a-zA-Z0-9]+...) are regular expression patterns designed for Gitleaks to identify secrets, not hardcoded credentials belonging to the skill or author.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation and use of Gitleaks through official GitHub and Docker Hub channels, which are trusted and well-known sources for security tooling.
[COMMAND_EXECUTION]: Shell commands provided for repository scanning and git history cleanup are standard security operations and follow best practices, such as using the --redact flag to protect output logs.
[PROMPT_INJECTION]: The skill processes data from external repositories, representing an ingestion surface for indirect prompt injection. However, its primary function is static analysis rather than natural language reasoning, and it includes support for allowlisting to manage data integrity. Ingestion points: Git repositories and source code files. Boundary markers: Gitleaks allowlist configuration and redaction features. Capability inventory: Execution of shell commands via Gitleaks and git. Sanitization: Use of predefined regex patterns and entropy thresholds.

secrets-gitleaks