drawio-roadmap-snake
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions require the agent to execute a local Python script, scripts/validate_snake_roadmap.py, to check the geometry and structure of Draw.io files. The script is provided in the skill package and performs static validation.
- [INDIRECT_PROMPT_INJECTION]: The skill operates on .drawio (XML) files which can contain untrusted labels and category text. Since the agent is instructed to read and inspect this content without sanitization or boundary markers, malicious instructions within a diagram could potentially influence the agent.
- Ingestion points: The agent reads content from .drawio files during the workflow and via the validation script.
- Boundary markers: No delimiters or ignore-instructions are used to separate diagram data from the agent's task instructions.
- Capability inventory: The agent can read and write files and execute local scripts using the python3 command.
- Sanitization: No sanitization is performed on the text items within the diagram files before they are processed by the agent.
- [DYNAMIC_EXECUTION]: The provided script scripts/validate_snake_roadmap.py uses the standard xml.etree.ElementTree library for parsing. This library is known to be vulnerable to XML-based attacks like Billion Laughs (DoS) or XXE if used to parse untrusted files.
Audit Metadata