readme-sync
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes content from SKILL.md files found in subdirectories. If these files contain malicious instructions within their description fields, those instructions would be echoed into the resulting README table. This creates a surface for indirect prompt injection if the generated output is subsequently processed by an AI agent. \n
- Ingestion points: The bash script reads the 'description' field from 'skills/*/SKILL.md'.\n
- Boundary markers: No markers are used to isolate the interpolated description data in the generated table row.\n
- Capability inventory: Shell execution via 'find', 'grep', 'sed', and 'echo' across all analyzed scripts.\n
- Sanitization: Basic length limiting (75 characters) and double-quote removal are performed, but no validation or escaping of the actual content is present.\n- [COMMAND_EXECUTION]: The process section provides a bash one-liner that uses 'find', 'grep', and 'sed' to iterate through the filesystem and extract metadata from subdirectories.\n- [EXTERNAL_DOWNLOADS]: The validation process utilizes 'bunx markdownlint-cli', which involves fetching and executing a third-party package from the npm registry.
Audit Metadata