vibe-coding-re-scan
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze untrusted codebase data to identify technical stacks, creating an attack surface where malicious content in scanned files could influence agent behavior.
- Ingestion points: The agent is directed to scan the user's project directory, including filenames, directory structures, and the contents of configuration files such as package.json, requirements.txt, manage.py, and Gemfile in SKILL.md.
- Boundary markers: No delimiters or specific instructions are provided to the agent to distinguish between legitimate codebase metadata and potential malicious instructions embedded within the scanned files.
- Capability inventory: The agent is tasked with counting files, identifying directory purposes, and writing a summary of its findings to a local file named progress.txt.
- Sanitization: The skill lacks instructions for sanitizing or validating the data extracted from the codebase before it is formatted into the scan output or saved to the progress log.
Audit Metadata