MCP server exposing AgentRove chat tools (create_chat, create_message, get_messages)
The AgentRove MCP server has three tools (create_chat, create_message, get_messages) with minimal tooling infrastructure visible in the provided source. The Dockerfile and pyproject.toml reveal backend infrastructure but do NOT contain the actual tool definitions, schemas, or parameter documentation. The repo reference (mcp-server/server.py) is mentioned but its contents are not provided in the source code sample. Without access to the actual tool registration code, parameter schemas, and descriptions, scoring is based on what can be inferred: all three tools appear to have only single-line descriptions ("Create a new chat session", "Create a message in a chat session", "Get messages from a chat session") with no visible input/output schemas, no parameter documentation, no error handling guidance, and no composition patterns. The server is STDIO-only (based on TRANSPORT field), which is a hard architecture limitation unrelated to definition quality but affects usability. Based on the extremely limited evidence available, none of the tools meet production quality baselines.
Create a new chat session
Create a message in a chat session
Get messages from a chat session
No input schemas visible for any tool. Cannot verify parameter names, types, descriptions, enums, constraints, or validation rules.
Tool descriptions are under 20 characters and generic. 'Create a new chat session' lacks WHEN to use it, prerequisites, side effects, or what it returns. Does not meet the 50-200 character baseline for LLM-optimized descriptions.
No output schema documentation visible. LLMs cannot know what fields to expect or how to chain results to downstream tools.
No error handling guidance. No indication of what errors are retryable, what should trigger user correction, or what recovery steps are available.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 25 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 33 | - | v1 |
Tool parameters are completely undocumented. Cannot determine which parameters are required, what types they accept, what values are valid, or what formats are expected.
No composition or chaining guidance. Cannot verify whether create_chat returns a chat_id that create_message accepts, or whether get_messages returns message_id for further operations.
Source code for mcp-server/server.py not provided in evaluation sample. Tool definitions are inferred from metadata only, not directly visible. Per hard scoring rules, each tool's score is capped at 50 when definitions are not directly visible.