open-design-integrations
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands
claude mcp listorcodex mcp listto determine the status of installed MCP servers. These are read-only commands used for configuration discovery within the environment. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the user's message to determine which integrations to display or configure.
- Ingestion points: The instruction
Request: the rest of the user's messageinSKILL.mdidentifies where external data enters the context. - Boundary markers: There are no explicit delimiters or boundary markers (like XML tags or triple quotes) used to isolate the user input from the rest of the prompt.
- Capability inventory: The skill has the capability to execute shell commands (
claude mcp list) and call integration tools (list_open_design_integrations). - Sanitization: No explicit sanitization of the user input is performed, although the instructions include a high-level policy to never ask for or write API keys.
Audit Metadata