ui-color-palette-generate-code
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes external base and themes configuration data, creating a potential surface for indirect prompt injection.
- Ingestion points: Data enters through the base and themes parameters in SKILL.md.
- Boundary markers: No specific delimiters are used for the input data, but the workflow instructions include a directive to "Do not read or analyze the returned code," which limits the agent's interaction with the output.
- Capability inventory: The skill utilizes the generate_code MCP tool and standard CLI tools (gh-cli, gitlab-cli-skills) for repository management.
- Sanitization: Input data does not appear to undergo explicit sanitization or validation within the instruction set.
Audit Metadata