Loading ToolBench
Spring Boot MCP server (HTTP) that bundles hexagonal bounded context tools (requirements, ubiquitous-language, use-cases, bounded-context, project, adr, actor-register, model-evaluation) behind a shared RDF dataset lifecycle
Critical gaps across all dimensions. No tool schemas visible in source code, only pom.xml files and configuration shown. Tool definitions are inferred from descriptions in the problem statement, not from actual code inspection. Cannot verify parameter types, input validation, or output structures. Descriptions are present but generic (10-50 chars), lacking LLM-optimized guidance on WHEN to use each tool or HOW to chain them. No evidence of error handling, recovery guidance, or permission gates. Spring AI @McpTool annotations mentioned but actual tool implementations not provided.
Add a new actor (HumanActor, SystemActor, LegalActor, or GroupActor) to the actor register
Retrieve an actor by identifier from the actor register
List all actors in the actor register
Update an existing actor in the actor register
Add a new architecture decision record with optional requirement and bounded context references
Retrieve an architecture decision record by identifier
List all architecture decision records
Set the status of an architecture decision record
No input schemas visible in source code. Tool definitions inferred from problem statement only; actual Spring AI @McpTool implementations not provided. Cannot verify parameter types, constraints, or validation.
Descriptions are generic and under 50 characters for most tools. Lack LLM-optimized guidance on WHEN to use each tool, prerequisites, or how to chain with other tools. E.g., 'Retrieve an actor by identifier' does not explain what an actor is or when to call actor_list vs actor_get.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | 2026-07-28+ | v2 |
Mark an architecture decision record as superseded by another
Add a new bounded context
Retrieve a bounded context by identifier
Link a term from the ubiquitous language to a bounded context
List all bounded contexts
Analyze the impact of changes across the model
Check for orphaned requirements, use cases, or bounded contexts with no cross-context references
Export the project model to a backup file
Retrieve a resource from the store by IRI
Generate a matrix of roles and their associated use cases
Validate the consistency and integrity of the RDF store
Generate an HTML report of the store contents and structure
Analyze term co-occurrence patterns across the ubiquitous language
Generate a traceability matrix across requirements, use cases, and bounded contexts
No evidence of error handling, recovery guidance, or permission gates in any tool. No indication of which tools are idempotent, retryable, or destructive. Agents cannot determine safe retry strategies.
No tool annotations visible (readOnlyHint, destructiveHint, idempotentHint). Agents cannot distinguish read-only discovery tools from write operations. All 22 tools appear unmarked.
No pagination, result limits, or output schema documentation visible. Tools like list_actors, trace_matrix, and store_overview could return unbounded results, risking context window exhaustion.