share-internal-doc
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: The skill incorporates robust privacy guidelines, instructing the agent to actively scrub shared documents of credentials, private local file paths (e.g., localhost, private directories), and sensitive personal information. It explicitly differentiates between technical identifiers and secret values to ensure only necessary data is shared.
- [INDIRECT_PROMPT_INJECTION]: While the skill processes external and internal documents (a typical surface for indirect prompt injection), it requires the agent to verify the recipient suitability and sanitizes the content according to strict organizational privacy rules. The inclusion of manual verification steps for links and citations further mitigates the risk of processing malicious instructions embedded in data.
- [EXTERNAL_DOWNLOADS]: The skill references established documentation from trusted and well-known sources (OpenAI, Substack, Hada.io) for editorial and prompting best practices. These references are used as informational guidelines rather than sources for executable code or automated package installations, which the skill explicitly prohibits.
Audit Metadata