indykite-mcp-server
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes and presents data retrieved from the remote IndyKite MCP server, such as tool definitions and resource descriptions, which could potentially be manipulated at the source to influence agent behavior.\n
- Ingestion points: Untrusted data enters the agent context through the output of
curlrequests performed by thescripts/mcp-call.shscript during discovery and tool execution.\n - Boundary markers: No specific delimiters or "ignore instructions" warnings are mentioned to the agent to isolate external data from its primary operating rules.\n
- Capability inventory: The skill allows the agent to execute network requests to the IndyKite platform, facilitating graph queries and authorization evaluations.\n
- Sanitization: The
scripts/mcp-call.shscript implements regex-based validation for JSON-RPC methods and resource names to prevent shell command injection.\n- [COMMAND_EXECUTION]: The skill uses a local bash script (scripts/mcp-call.sh) to executecurlcommands for API communication. This is the primary function of the skill and is implemented with input validation for key parameters to mitigate injection risks.
Audit Metadata