share-example
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data including
transcript.jsonland workspace files which could potentially contain embedded instructions if they originated from untrusted sources. - Ingestion points: Processes logs from
<appDataDir>/brain/and project files from the local workspace. - Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are applied to the content of the files during processing.
- Capability inventory: Reads local files, writes to the
./share_post/directory, and generates a pre-filled URL for GitHub Discussions. - Sanitization: Includes explicit instructions to redact sensitive API key patterns (e.g.,
AIza,GEMINI_API_KEY,sk-,Bearer) and anonymize local directory paths (replacing/Users/<name>with~). - [DATA_EXFILTRATION]: The skill's primary purpose is to package local data for external sharing on GitHub. However, this is done through a 'Show and tell' pre-filled URL and local file staging, requiring the user to manually drag-and-drop files and submit the discussion post, ensuring human-in-the-loop control. The target repository belongs to the skill's author organization.
Audit Metadata