Azex MCP Server — AI agents interact with Azex via the Model Context Protocol
Azex MCP server has 13 tools with basic naming conventions (verb_noun pattern present) but critical gaps in schema visibility and parameter documentation. Tool descriptions are present but generic (10-50 chars, well below the 50-200 char LLM-optimized baseline). Source code shows tool registration in server.ts but actual input schemas and parameter descriptions are not visible in the provided code excerpt, only tool names and brief descriptions are shown. This forces conservative scoring: cannot verify schema completeness, parameter types, or output documentation. Error handling is present (formatError function with status-specific messages) but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite risk classifications being declared in the metadata.
Create a new API key
Create a chat completion using the Azex API
Create embeddings using the Azex API
Create a message using the Azex API
Get current account information
Get account balance information
Get deposit address and information
Get usage statistics
Input schemas not visible in source code. Tool registration in server.ts shows tools are registered with inputSchema property, but actual schema definitions (Zod types, JSON Schema) are not provided in the code excerpt. Cannot verify parameter types, constraints, or descriptions.
Tool descriptions are extremely brief (10-50 characters). Baseline for LLM-optimized descriptions is 50-200 chars. Current descriptions lack context on WHEN to use each tool, prerequisites, or what the tool returns. Example: 'Get current account information' does not explain what fields are returned or when to call it vs get_balance.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 32 | <=2025-11-25 | v2 |
List API keys
List available models
List account transactions
Trigger a blockchain scan for deposit address
Revoke an API key
No tool annotations despite risk classifications. Metadata declares risk levels (READ_ONLY, WRITE, DESTRUCTIVE) but tools are not registered with readOnlyHint, destructiveHint, or idempotentHint annotations. This prevents agents from understanding which tools are safe to retry and which have irreversible consequences.
Output schemas not documented. No evidence in source that tool responses document their structure (fields, types, pagination). LLMs cannot plan downstream tool calls or extract required data without knowing what fields to expect.
Error handling lacks recovery guidance. formatError() function returns status-specific messages but does not guide the agent on next steps. Example: 'Not found: ...' does not suggest calling a search tool or listing available options.