🌐 Website | 📚 Document

Acontext is the memory stack for production AI agents. Think of it as Supabase for agent memory.

Unifies short-term memory, mid-term state, and long-term skill for production AI agents.

• Context data is scattered — messages, files, and skills live in different storages with no unified interface
• No observability on agent state — you can't track success rates, replay trajectories, or know if your agent is actually working
• Your agent's memory is a black box — vector stores and key-value memory are opaque, not inspectable, and not version controllable

• Short-term Memory — unified storage for messages, files, and artifacts — integrated with Claude Agent SDK, AI-SDK, OpenAI SDK...
• Mid-term State — replay trajectories, track success rates, and monitor agents in real-time
• Long-term Skill — agents distill successful/failed task outcomes into reusable, human-readable skill files, improving with every run

• Short-term Memory
  • Session: save agent history from any LLM, any modality
• Mid-term State
  • State Tracking: collect agent tasks and results in near real-time
• Long-term Skill
  • Skill Memory - agents automatically build and update skills from successful/failed sessions

1. Go to Acontext.io, claim your free credits.
2. Go through a one-click onboarding to get your API Key (starts with sk-ac)

curl -fsSL https://install.acontext.io | sh

mkdir acontext_server && cd acontext_server
acontext server up

We're maintaining Python and Typescript SDKs. The snippets below are using Python.

Click the doc link to see TS SDK Quickstart.

pip install acontext

import os
from acontext import AcontextClient

# For cloud:
client = AcontextClient(
    api_key=os.getenv("ACONTEXT_API_KEY"),
)

# For self-hosted:
client = AcontextClient(
    base_url="http://localhost:8029/api/v1",
    api_key="sk-ac-your-root-api-bearer-token",
)

Store a message, get agent state, and retrieve learned skills — one API for each layer.

session = client.sessions.create()
space = client.learning_spaces.create()
client.learning_spaces.learn(space.id, session_id=session.id)

# 1. Short-term Memory — store messages in any LLM format
client.sessions.store_message(
    session_id=session.id,
    blob={"role": "user", "content": "Deploy the new API to staging"},
)
# ... your agent runs ...
msgs = client.sessions.get_messages(session_id=session.id)

# 2. Mid-term State — flush to trigger processing, then get state
client.sessions.flush(session.id)
summary = client.sessions.get_session_summary(session_id=session.id)
print(summary)

# 3. Long-term Skill — wait for learning, then retrieve skills
client.learning_spaces.wait_for_learning(space.id, session_id=session.id)
skills = client.learning_spaces.list_skills(space.id)
for skill in skills:
    print(f"{skill.name}: {skill.description}")

flush and wait_for_learning are blocking helpers for demo purposes. In production, task extraction and learning run in the background automatically — your agent never waits.

• Context Engineering — Compress context with summaries and edit strategies
• Disk — Virtual, persistent filesystem for agents
• Sandbox — Isolated code execution with bash, Python, and mountable skills
• Agent Tools — Disk tools, sandbox tools, and skill tools for LLM function calling

Download end-to-end scripts with acontext:

Python

acontext create my-proj --template-path "python/openai-basic"

More examples on Python:

• python/openai-agent-basic: openai agent sdk template
• python/openai-agent-artifacts: agent can edit and download artifacts
• python/claude-agent-sdk: claude agent sdk with ClaudeAgentStorage
• python/agno-basic: agno framework template
• python/smolagents-basic: smolagents (huggingface) template
• python/interactive-agent-skill: interactive sandbox with mountable agent skills

Typescript

acontext create my-proj --template-path "typescript/openai-basic"

More examples on Typescript:

• typescript/vercel-ai-basic: agent in @vercel/ai-sdk
• typescript/claude-agent-sdk: claude agent sdk with ClaudeAgentStorage
• typescript/interactive-agent-skill: interactive sandbox with mountable agent skills

To learn more about long-term skill and what Acontext can do, visit our docs or start with What is Long-term Skill?

Star Acontext on Github to support and receive instant notifications

graph TB
    subgraph "Client Layer"
        PY["pip install acontext"]
        TS["npm i @acontext/acontext"]
    end
    
    subgraph "Acontext Backend"
      subgraph " "
          API["API<br/>localhost:8029"]
          CORE["Core"]
          API -->|FastAPI & MQ| CORE
      end
      
      subgraph " "
          Infrastructure["Infrastructures"]
          PG["PostgreSQL"]
          S3["S3"]
          REDIS["Redis"]
          MQ["RabbitMQ"]
      end
    end
    
    subgraph "Dashboard"
        UI["Web Dashboard<br/>localhost:3000"]
    end
    
    PY -->|RESTFUL API| API
    TS -->|RESTFUL API| API
    UI -->|RESTFUL API| API
    API --> Infrastructure
    CORE --> Infrastructure

    Infrastructure --> PG
    Infrastructure --> S3
    Infrastructure --> REDIS
    Infrastructure --> MQ
    
    
    style PY fill:#3776ab,stroke:#fff,stroke-width:2px,color:#fff
    style TS fill:#3178c6,stroke:#fff,stroke-width:2px,color:#fff
    style API fill:#00add8,stroke:#fff,stroke-width:2px,color:#fff
    style CORE fill:#ffd43b,stroke:#333,stroke-width:2px,color:#333
    style UI fill:#000,stroke:#fff,stroke-width:2px,color:#fff
    style PG fill:#336791,stroke:#fff,stroke-width:2px,color:#fff
    style S3 fill:#ff9900,stroke:#fff,stroke-width:2px,color:#fff
    style REDIS fill:#dc382d,stroke:#fff,stroke-width:2px,color:#fff
    style MQ fill:#ff6600,stroke:#fff,stroke-width:2px,color:#fff

Join the community for support and discussions:

• Discuss with Builders on Acontext Discord 👻
• Follow Acontext on X 𝕏

• Check our roadmap.md first.
• Read contributing.md

[![Made with Acontext](https://assets.memodb.io/Acontext/badge-made-with-acontext.svg)](https://acontext.io)

[![Made with Acontext](https://assets.memodb.io/Acontext/badge-made-with-acontext-dark.svg)](https://acontext.io)

This project is currently licensed under Apache License 2.0.