The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation instructions direct users to clone a repository from an external source (github.com/elzobrito/ESAA-Security.git) which is not categorized as a trusted organization or well-known service.
[REMOTE_CODE_EXECUTION]: The skill instructs the user to clone an external repository and then execute Python scripts from within that repository (python orchestrator.py). This sequence constitutes the download and execution of code from an unverified remote source.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted code from a target repository (AUDIT_TARGET_REPO) for security analysis. This creates a surface for indirect prompt injection where instructions embedded in the audited code could influence the agent's behavior.
Ingestion points: Files within the user-specified AUDIT_TARGET_REPO (e.g., .py, .js, .java, .java).
Boundary markers: The documentation references schema-validated outputs and evidence requirements, which serve as boundary controls.
Capability inventory: The agent requires file read access to the target repository, write access to a reports/ directory, and the ability to execute security checks.
Sanitization: The skill identifies "user_input_sanitization" and "schema-validated outputs" as internal mechanisms for maintaining audit hygiene.

esaa-security-audit