OdinClaw is an agentic AI gateway that gives your applications web search, code execution, file analysis, and image generation — all through a single OpenAI-compatible API. Built and operated by Odin Labs.
OdinClaw is fully OpenAI-compatible. Switch your existing apps to use agentic tools without rewriting a single function.
base_url = "https://api.openai.com/v1"
api_key = "sk-..."base_url = "api.claw.odin-labs.ai/v1"
api_key = "oc_live_..."Works with any OpenAI SDK — Python, Node.js, Go, Ruby, and more.
Every OdinClaw model can use tools automatically. No extra configuration, no separate infrastructure — tools are available out of the box.
Your AI can search the web in real time. No stale training data — get live results, news, and factual answers on demand.
Pass any URL and the model receives the full page content. Perfect for research, summarization, and content-aware responses.
Execute Python code securely in an isolated environment. Run calculations, process data, test algorithms — without leaving the chat.
Upload PDFs, spreadsheets, images, or code files and let the model analyze, extract, and reason over them natively.
Generate images directly from natural language prompts. Integrate visual creation into your AI workflows without a separate API.
Our MCP servers work with any MCP-compatible client — Claude Desktop, Cursor, VS Code, and more. Install via npm, configure your environment, and your AI has new capabilities.
Persistent AI memory across sessions.
npm install
@odinlabs-ai/brain-mcp-server@1.0.0
Enterprise governance for AI decisions.
npm install
@odinlabs-ai/governance-mcp-server@1.0.0
AI-powered sales automation.
npm install
@odinlabs-ai/sales-mcp-server@1.0.0
Intelligent learning experiences.
npm install
@odinlabs-ai/academy-mcp-server@1.0.0
All servers are published on GitHub Packages under @odinlabs-ai.
Security is enforced at the architecture level, not the configuration level. Every request is validated, classified, and logged before it reaches a model.
Every request passes through two independent harnesses — gateway checks AND governance validation.
14 patterns detect sensitive content across 4 severity levels before it leaves your system.
Instantly disable specific capabilities (LLM, code-gen, web-fetch) per-target with full audit trail.
SHA-256 hash chain creates tamper-evident logs of every AI interaction.
Five service capabilities available via standard REST endpoints at /ap/v1/*. Send a task, get a structured result. No SDK required.
POST /ap/v1/agent/tasks
Content-Type: application/json
Authorization: Bearer oc_live_...
{
"capability": "code-review",
"target": "https://github.com/org/repo/pull/42",
"options": {
"model": "claude-sonnet-4-6",
"include_suggestions": true
}
}Access models from Anthropic, OpenAI, Google, Meta, and more — all with the same OpenAI-compatible interface and unified billing.
No credit card required. No approval process. Just sign up and start building.
Everything you need to get started
For individuals building AI applications
For teams shipping production AI features
For high-volume production workloads
All plans include all models and all tools. Token overages billed at transparent per-token rates.
No credit card. No approval process. Start building with agentic AI today.
app.claw.odin-labs.ai