The Agent Skills Directory

[SAFE]: The skill instructions are primarily defensive, emphasizing the importance of deterministic validation, manifest integrity, and secure installation procedures. It explicitly rejects insecure practices like piping remote content directly to a shell.
[PROMPT_INJECTION]: The skill handles external data including contracts, scoring inputs, and bundles, which constitutes an indirect prompt injection surface. Evidence Chain: 1. Ingestion points: Validation of external contracts, bundles, and manifests as described in SKILL.md. 2. Boundary markers: The skill instructs the agent to validate configured root boundaries. 3. Capability inventory: The skill is authorized to perform file system operations, specifically uninstalls based on manifests. 4. Sanitization: The skill enforces SHA-256 checksum verification for external archives and requires manifest matching before deletions.

ads-validate