docs-generator
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from upstream security analysis tasks, which could theoretically contain malicious instructions designed to influence report output.
- Ingestion points: Ingests analysis conclusions, logs, and findings from previous task stages.
- Boundary markers: Explicitly instructs the agent to replace sensitive information (tokens, passwords) with placeholders, which helps prevent accidental data exposure.
- Capability inventory: Limited to writing Markdown files in the project directory and calling internal diagramming tools; no network or arbitrary code execution capabilities are present.
- Sanitization: Enforces a strict Evidence-Finding-Path chain to ensure all documentation is grounded in verifiable steps.
- [SAFE]: The skill references industry-standard tools (IDA Pro, radare2, Frida) and adopts structural layouts from established security vendors (e.g., Kaspersky, Huorong) solely for formatting and professional documentation purposes. No external code is downloaded or executed.
Audit Metadata