Open-source MCP server for AAS (Asset Administration Shell) APIs using OpenAPI specifications. Dynamically generates MCP tools from OpenAPI specs with safety-focused curation (allowlist filtering, read-only by default, operation ID aliasing, limit parameter capping).
Single tool 'dynamic_openapi_tools' has a generic description (89 chars) that lacks specificity about what operations are exposed, when to use it, or what it returns. No input schema visible in source code, tool definition appears inferred from OpenAPI spec curation logic rather than explicitly registered with MCP schema. Tool name is a noun phrase ('dynamic_openapi_tools') rather than action verb (should be 'call_openapi_operation' or similar). Curation logic shows safety-focused design (allowlist, read-only default) but this is implementation detail, not reflected in tool interface. No parameter descriptions, no output schema documentation, no error handling guidance visible.
Dynamically generated tools from OpenAPI specification paths. Tools are created from curated OpenAPI operations (GET /shells and other allowlisted endpoints). Each tool corresponds to an OpenAPI operation with input parameters from the operation's schema and output from the response schema.
Tool definition inferred from OpenAPI curation, not explicitly registered with MCP schema. No input schema visible in source code.
Tool name is noun phrase 'dynamic_openapi_tools' instead of action verb. Should be 'call_openapi_operation', 'invoke_aas_api', or similar to signal intent.
Tool description (89 chars) is generic and lacks specificity. Does not explain which operations are exposed, when to call this vs other tools, or what structure is returned. Missing: operation list, parameter guidance, output format.
No input parameters documented. LLM cannot determine how to invoke the tool or what operation/parameters to pass. Curation allowlist and aliases are server-side config, not exposed to agent.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 29 | <=2025-11-25 | v2 |
No output schema documented. LLM cannot plan downstream tool calls or extract required fields (e.g., shell IDs for chaining). Response structure is opaque.