The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions and references contain shell commands using git, grep, find, and xargs to audit repository source code and history for secrets and configuration errors.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of third-party tools like gitleaks, trufflehog, and pre-commit from trusted registries and official GitHub repositories.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection because it analyzes untrusted content from repository files, commit messages, and documentation.
Ingestion points: Untrusted data is ingested from the git object store and repository files via git grep and git log.
Boundary markers: No specific delimiters or boundary instructions are used to separate untrusted repository content from instructions.
Capability inventory: The skill performs subprocess execution using git and various security-focused CLI tools.
Sanitization: No explicit sanitization or validation of audited data is mentioned before incorporating findings into the agent context.

repo-sentinel