lov-skill-pricing

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze third-party Skill source code, README files, and capability briefs. This represents an indirect prompt injection surface. However, the workflow (specifically Step 2: Separate capability from internal context) implements a security boundary by instructing the agent to identify and exclude private context—such as keys, internal paths, and unverified promises—from the public-facing output. The risk is assessed as low due to these explicit constraints.
  • [COMMAND_EXECUTION]: The skill documentation references a local Python script (scripts/validate_skill.py) for source validation. A technical audit of the script confirms it is a standard linting utility that uses yaml.safe_load for secure parsing and does not contain any network operations, subprocess calls, or persistence mechanisms.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 10:09 PM
Security Audit — agent-trust-hub — lov-skill-pricing