audit-log
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided descriptions of business processes and existing trackers to generate audit log definitions.
- Ingestion points: Step 2 captures user input regarding business events, purposes, and current processes (e.g., "our process is...", "it is in a sheet").
- Boundary markers: The instructions mandate keeping prose separate from machine-readable text and recommend UTF-8 with byte order marks for CSV outputs to maintain data integrity.
- Capability inventory: The skill has the capability to generate SQL DDL, JSON Schema, and CSV templates based on processed user input.
- Sanitization: The skill contains explicit safety notes prohibiting the use of real names, salaries, medical, or banking data, requiring synthetic placeholders for any examples.
- [EXTERNAL_DOWNLOADS]: The skill references external documentation and module catalogs located on GitHub.
- Evidence: The README and metadata link to the vendor's GitHub repository at
github.com/sickn33/agentic-awesome-skillsfor additional module references. - [SAFE]: The skill contains clear security boundaries regarding tool usage and credential handling.
- Evidence: Instructions explicitly state that a user saying "connected" is not evidence of tool access and strictly forbid the agent from asking for Notion passwords or tokens.
Audit Metadata