knowledge-doc-writing
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external, potentially untrusted data to generate documentation.
- Ingestion points: Processes conversation transcripts, raw source material (official docs, specs, meeting notes), and existing documents provided by the user (SKILL.md, S1).
- Boundary markers: The instructions do not specify the use of clear delimiters or instructions to ignore embedded commands within the ingested text.
- Capability inventory: The skill possesses capabilities for text generation, Mermaid diagramming, and HTML/SVG generation. It does not have hardcoded capabilities for network operations or sensitive file system access.
- Sanitization: No explicit sanitization or filtering of input data is defined before the content is transformed into documentation sections.
- [COMMAND_EXECUTION]: The skill's reference files (e.g., references/blocks.md) contain numerous examples of CLI commands, including OpenShift (oc), Podman, and shell scripts.
- Evidence: These commands are explicitly labeled as documentation templates and examples (e.g., '最小可運行範例' / 'minimal runnable example') for inclusion in the generated output, not for execution by the agent environment itself.
- [EXTERNAL_DOWNLOADS]: The skill refers to several other AI agent skills for specialized tasks.
- Evidence: It suggests invoking 'humanizer-zh' for language polishing and 'infographic-design' for SVG diagram styling if they are available in the agent's environment. These are documented as optional orchestration steps rather than remote code execution vulnerabilities.
Audit Metadata