Kwami exposes 3 tools with critical definition gaps. All three tools have minimal descriptions (under 50 chars) and vague, generic names that violate verb_noun conventions. The 'execute' and 'callTool' tools are nearly identical, creating confusion. Input schemas are present but lack parameter descriptions, a hard requirement. No output schemas are documented. The tools appear to be meta-tools for invoking other tools rather than domain-specific capabilities, which is an anti-pattern for agent composition.
Call a tool on an MCP server with parameters
Execute a tool by name with provided parameters
List all available tools from an MCP server
Duplicate meta-tools: 'execute' and 'callTool' are nearly identical, both accepting 'name' and 'params'. This violates single-responsibility and creates LLM confusion about which to use.
All parameter descriptions missing. 'name' and 'params' fields have generic descriptions ('The name of the tool to execute', 'Parameters to pass to the tool handler') that do not explain constraints, format, or valid values. LLMs cannot infer what tool names are valid or what structure 'params' should have.
No output schemas documented. Tools return 'unknown' type with no guidance on structure, fields, or error responses. Agents cannot plan downstream calls or extract data reliably.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 22 | <=2025-11-25 | v2 |
Generic tool names violate verb_noun convention. 'execute' and 'callTool' are vague meta-operations; 'listTools' is discovery-only. No domain-specific tools (e.g., 'create_task', 'send_message') are exposed, limiting agent utility.
Tool descriptions are under 50 characters and lack context. 'Execute a tool by name with provided parameters' does not explain when to use this vs 'callTool', what happens on error, or what the return value contains.