Machine-readable contracts for building AI agents on top of bitmex-cli.

tool-catalog.json is the canonical command contract.

• Coverage: 134 commands with parameters, types, defaults, and examples
• Every command includes: group, description, auth requirement, dangerous flag, and parameter schemas
• 32 commands are marked dangerous: true (orders, withdrawals, transfers, cancel-all, staking)

error-catalog.json is the error routing contract.

• 9 error categories with retry/backoff guidance

import json

with open("agents/tool-catalog.json") as f:
    catalog = json.load(f)

for cmd in catalog["commands"]:
    print(f"{cmd['group']}/{cmd['name']}: {cmd['description']}")
    if cmd.get("dangerous"):
        print("  ⚠ requires human confirmation")

The catalog contains everything needed to generate tool schemas for OpenAI, Anthropic, MCP, LangChain, or any other framework. Each command entry includes:

{
  "name": "order-buy",
  "group": "trade",
  "command": "bitmex order buy <PAIR> <VOLUME>",
  "description": "Place a buy order.",
  "auth_required": true,
  "dangerous": true,
  "parameters": [
    { "name": "PAIR", "type": "string", "required": true, "positional": true },
    { "name": "--type", "type": "string", "required": false }
  ],
  "example": "bitmex order buy BTCUSD 0.001 --type limit --price 50000 --validate"
}

safe_commands = [c for c in catalog["commands"] if not c.get("dangerous")]
dangerous_commands = [c for c in catalog["commands"] if c.get("dangerous")]

• ../CONTEXT.md: Runtime context for tool-using agents
• ../AGENTS.md: Full integration guide
• ../skills/: Goal-oriented workflow skills
• ../gemini-extension.json: Gemini CLI extension manifest