understand-onboard
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from
.understand-anything/knowledge-graph.json. If the analyzed project contains malicious filenames, code summaries, or metadata designed to influence the LLM, these could be executed as instructions during the guide generation process. - Ingestion points: Reads project metadata, nodes, layers, and tours from
.understand-anything/knowledge-graph.json(SKILL.md, Instructions 2-5). - Boundary markers: None; the skill does not use delimiters or instructions to ignore embedded commands within the ingested JSON data.
- Capability inventory: The skill can write the resulting guide to the file system at
docs/ONBOARDING.md(SKILL.md, Instruction 9). - Sanitization: No sanitization, escaping, or validation of the graph content is performed before it is interpolated into the final output.
Audit Metadata