umbraco-skill-validator
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local TypeScript script (
validate-links.ts) using thenpx tsxcommand to perform deterministic validation of repository files. - [EXTERNAL_DOWNLOADS]: The skill downloads and installs standard Node.js packages (
cockatiel,glob,tsx) from the NPM registry to support its validation logic. - [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface when processing repository content.
- Ingestion points: The script reads the contents of all
SKILL.mdfiles in the repository to extract URLs and references for checking. - Boundary markers: There are no explicit boundary markers or instructions to ignore embedded malicious content when the extracted data is formatted into a report for the fixer subagent.
- Capability inventory: The skill possesses powerful capabilities including
Edit,Bash, andTasktools, which are used by the subagent to suggest and apply changes based on the validation report. - Sanitization: The validation script does not sanitize or filter the content of extracted links or paths before including them in the data passed to the AI fixer agent.
Audit Metadata