SDK Documentation
Anchor provides official SDKs for Python and TypeScript/JavaScript to make integration easier. All SDKs provide full API coverage with type safety, error handling, and framework integrations.
Full-featured Python SDK with support for LangChain, CrewAI, and Mem0 integrations.
pip install anchoraiTypeScript/JavaScript SDK with full type definitions and modern async/await support.
npm install anchoraiSDK Structure
Both SDKs are organized around 5 core namespaces:
- agents - Agent registry and lifecycle management
- config - Versioned agent configuration
- data - Governed key-value data storage
- checkpoints - State snapshots and rollback
- audit - Hash-chained audit trail
Quick Start
Get started with Anchor in just a few lines of code:
Python
from anchor import Anchor
anchor = Anchor(api_key="your-api-key")
agent = anchor.agents.create("support-bot")
anchor.config.update(agent.id, {"policies": {"block_pii": True}})
result = anchor.data.write(agent.id, "key", "value")
print(result.allowed) # TrueTypeScript
import { Anchor } from 'anchorai';
const anchor = new Anchor({ apiKey: 'your-api-key' });
const agent = await anchor.agents.create('support-bot');
await anchor.config.update(agent.id, { policies: { block_pii: true } });
const result = await anchor.data.write(agent.id, 'key', 'value');
console.log(result.allowed); // trueFramework Integrations
Anchor integrates with popular AI frameworks to add governance to their memory systems:
- LangChain - Policy-checked memory for LangChain chains and agents
- CrewAI - Policy-checked shared memory for CrewAI crews
- Mem0 - Policy-checked memory operations with Mem0
How it works: Anchor works above memory frameworks. Memory frameworks handle "what does the agent remember?" Anchor handles "what is the agent allowed to remember, and can you prove it?" This adds policy enforcement, audit trails, and checkpoints without replacing the memory framework.
See the individual SDK documentation for integration examples.