kernel-marketplace
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local build script:
python scripts/daodan_build.py --check --support. This grants the agent the ability to run shell commands to validate and build kernel packages. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from various source files that could be manipulated to influence agent behavior.
- Ingestion points: The agent reads content from
plugin.toml,roles/,workflows/,skills/,contracts/,policies/, andadapters/directories. - Boundary markers: No specific delimiters or safety instructions are provided to the agent to treat the content of these files as passive data rather than executable instructions.
- Capability inventory: The agent is authorized to execute shell commands (specifically the
daodan_build.pyscript) and perform file system operations (staging source and generated results). - Sanitization: There is no evidence of input validation or sanitization for the natural language behavior or metadata embedded in the processed TOML and Markdown files.
Audit Metadata