AI Developer Relay - Centralized MCP server config and AI rules manager
aidrelay exposes 7 server management tools with consistent naming (verb_noun pattern: servers:list, servers:get, servers:create, etc.) and reasonable descriptions (avg ~80 chars). All tools have input schemas with type definitions and parameter descriptions. However, output schemas are not documented in the source code, parameter descriptions lack detail on constraints/formats, and error handling guidance is absent. No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels. Schemas are present but minimal, most parameters lack examples or validation rules. This is solid foundational work but falls short of production-grade agent tooling.
Creates a new server record and returns the persisted entity
Deletes a server record by ID
Finds a single server by its UUID
Imports MCP server configurations from detected client installations
Returns all server records ordered by name
Tests a stdio-transport MCP server by spawning it and sending an initialize JSON-RPC request over stdin/stdout. For sse/http servers it issues a GET request to the configured URL
Applies a partial update to an existing server record
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields (e.g., server UUID for chaining). Responses must include all IDs needed for follow-up operations.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels. servers:delete is DESTRUCTIVE, servers:create/update are WRITE, servers:list/get/test are READ_ONLY. Agents need this metadata to reason about safety.
Parameter descriptions lack validation rules and format constraints. E.g., 'type' param accepts 'stdio, sse, or http' but this is buried in description text, not formalized as enum. 'secretEnvKeys' and 'secretHeaderKeys' lack guidance on what constitutes a valid key name.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 66 | 2026-07-28+ | v2 |
No error handling guidance. Tools lack recovery hints (e.g., 'Server not found. Call servers:list to see available servers.'). Agents cannot self-correct on failures.
servers:import-from-clients has empty input schema ({}). No parameters documented. Unclear what this tool does or what preconditions must be met. Description is vague about which client installations are detected.