better-skill-review
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted skill directories, including their instructions (SKILL.md) and scripts. This creates a surface where a malicious target skill could attempt to perform indirect prompt injection to influence the reviewing agent's assessment or behavior.
- Ingestion points: The skill reads the entire content of a user-provided directory, specifically the SKILL.md, scripts/, and references/ folders during the linting and profiling steps.
- Boundary markers: The skill lacks explicit sanitization or delimiters for the untrusted content, though it mitigates risks by using a structured multi-agent workflow where a separate 'verification subagent' must independently validate findings.
- Capability inventory: The skill executes local Python and Bash scripts (validate.py, analyze.sh) and has the capability to dispatch general-purpose subagents for task delegation.
- Sanitization: No sanitization is performed on the ingested text before it is analyzed by the agent, which is a standard risk for auditing tools.
Audit Metadata