crowvault

Python SDK for CrowVault — MCP-native SDK for 327 AI dev tools — code generation via Model Context Protocol. Zero dependencies.

Install

pip install crowvault

Quick Start

from crowvault import CrowVaultClient

client = CrowVaultClient(api_key="cv_your_key_here")

# Generate a database schema
result = client.call("database-mcp", "design_schema",
                     aggregate="Order", database="postgresql")
print(result.text)

# Generate a Dockerfile
result = client.call("devops-mcp", "generate_dockerfile",
                     runtime="node", version="20")
print(result.text)

Authentication

# Option 1: Pass directly
client = CrowVaultClient(api_key="cv_...")

# Option 2: Environment variable
# export CROWVAULT_API_KEY=cv_...
client = CrowVaultClient()

API Reference

Call a Tool

result = client.call("database-mcp", "design_schema",
                     aggregate="Order",
                     database="postgresql",
                     outputFormat="prisma")

print(result.text)       # Generated code
print(result.success)    # True/False
print(result.meta)       # {server, tool, durationMs, ...}

Call with Dict Args

Use call_raw() when arg names conflict with Python keywords:

result = client.call_raw("api-mcp", "generate_openapi_spec", {
    "service": "users",
    "version": "v1",
    "format": "yaml",   # "format" is a Python builtin
})

List Servers & Tools

# All servers
servers = client.servers()
for s in servers:
    print(f"{s['name']} — {s['toolCount']} tools")

# All tools (or filter)
tools = client.tools(server="database-mcp")
tools = client.tools(search="schema")

Batch Execution

Execute up to 10 tools in parallel:

result = client.batch([
    {"server": "database-mcp", "tool": "design_schema",
     "args": {"aggregate": "Order", "database": "postgresql"}},
    {"server": "devops-mcp", "tool": "generate_dockerfile",
     "args": {"runtime": "node", "version": "20"}},
    {"server": "devops-mcp", "tool": "generate_k8s_deployment",
     "args": {"name": "order-service"}},
])

print(f"{result.succeeded}/{len(result.results)} succeeded")
for r in result.results:
    print(r.text[:100])

Workflows

Run multi-step pipelines that chain tools together:

# List available workflows
workflows = client.workflows()
for wf in workflows:
    print(f"{wf['id']} — {wf['stepCount']} steps")

# Run a workflow
result = client.run_workflow("microservice-scaffold",
                             name="orders",
                             language="node",
                             database="postgresql")

print(result)  # Step-by-step summary
print(result.success)
print(result.total_duration_ms)

for step in result.results:
    print(f"{step['tool']} — {step['durationMs']}ms")

Available workflows:

• microservice-scaffold — Microservice + schema + migration + Docker + K8s (5 steps)
• api-full-stack — OpenAPI + endpoint + tests + contracts (4 steps)
• ddd-complete — Bounded context + model + aggregate + entity + repo + schema (6 steps)
• event-driven — Kafka + event handler + DLQ (3 steps)
• deploy-stack — Dockerfile + Compose + K8s + Helm (4 steps)
• database-setup — Schema + ORM + migration + seed (4 steps)

9 Servers, 327 Tools

Error Handling

from crowvault import CrowVaultClient
from crowvault.client import CrowVaultError

client = CrowVaultClient()

try:
    result = client.call("database-mcp", "design_schema",
                         aggregate="Order", database="postgresql")
    print(result.text)
except CrowVaultError as e:
    print(f"Error: {e} (HTTP {e.status})")

Pricing

• Free: 50 calls/month, 2 API keys — Sign up free
• Developer: $49/month, 1,000 calls, 5 keys
• Team: $199/month, 10,000 calls, 10 keys
• Enterprise: Custom

License

MIT. Copyright TechSynergy Corp.