CrewAI tools for instanode.dev. Let a CrewAI agent provision a real Postgres database (pgvector pre-installed) or a webhook receiver mid-task with a single tool call — no signup, no Docker, no setup. Free anonymous tier (24h TTL). Prefer MCP? Use the instanode MCP server (Claude Code / Cursor / Copilot / Windsurf) instead.
pip install crewai-instanode-toolsfrom crewai import Agent, Task, Crew
from crewai_instanode_tools import (
ProvisionPostgresTool,
ProvisionWebhookTool,
ListResourcesTool,
)
provisioner = Agent(
role="Infrastructure provisioner",
goal="Give the team working database URLs with zero setup.",
backstory="Spins up Postgres + webhooks via instanode.dev.",
tools=[ProvisionPostgresTool(), ProvisionWebhookTool(), ListResourcesTool()],
)
task = Task(
description="Stand up a Postgres DB for the embeddings pipeline and report its DSN.",
agent=provisioner,
)
Crew(agents=[provisioner], tasks=[task]).kickoff()ProvisionPostgresTool—postgres://DSN, pgvector pre-installed.ProvisionWebhookTool— HTTPS receiver URL (stores recent request bodies).ListResourcesTool— enumerate resources owned by the current API key.
MongoDB, Redis/cache, NATS queue, and heartbeat-monitor tools are on the
roadmap, gated on the matching backend endpoints landing. They live on
the feature/full-api branch.
Set INSTANODE_API_KEY in your environment, or pass explicitly:
ProvisionPostgresTool(api_key="sk_...")| Tier | Postgres | Webhooks | Persistence |
|---|---|---|---|
| Anonymous (no key) | 10 MB / 2 connections | 100 stored | 24 hours |
| Paid | 500 MB / 5 connections | 1000 stored | Permanent |
- Python SDK: https://pypi.org/project/instanode/
- LangChain variant: https://pypi.org/project/langchain-instanode/
- LlamaIndex variant: https://pypi.org/project/llama-index-tools-instanode/
- MCP server (Claude Code / Cursor): https://www.npmjs.com/package/@instanode/mcp
- HTTP API: https://instanode.dev/llms.txt
MIT.