validate-skill
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- EXTERNAL_DOWNLOADS (MEDIUM): The skill uses
bunx @plaited/development-skillsto fetch code from the npm registry. The@plaitedorganization is not a recognized trusted source, making the dependency unverifiable. - REMOTE_CODE_EXECUTION (HIGH): The use of
bunxto execute an unverified third-party package allows for arbitrary code execution on the host machine at runtime. - INDIRECT PROMPT_INJECTION (HIGH): The skill has a high-risk attack surface as it is intended to process untrusted data (specifically for "Reviewing PRs").
- Ingestion points: The
[paths...]argument inSKILL.mdallows the agent to ingest arbitrary directories. - Boundary markers: None identified. There are no instructions to the agent to ignore embedded commands or instructions within the validated files.
- Capability inventory: The skill has
Bashaccess and executes code viabunx(File:SKILL.md). - Sanitization: No evidence of input validation or sanitization is present in the skill's instructions.
Recommendations
- AI detected serious security threats
Audit Metadata