Skip to content
Get packs

Hindsight

Open-source memory system for AI agents that aims to help them learn over time, not just recall chat history.

GitHub Download v1.1.0
Stars
57.9k
Forks
7.1k
Licence
MIT
Language
Python
Last update
4 days ago
Type
AI model / agent

What it is

Hindsight is an agent memory system that gives AI agents a place to store and search what they have learned. The README says most memory systems focus on recalling conversation history, while Hindsight aims to make agents that learn over time. It offers retain, recall and reflect operations, and can run as a server with a Python, Node.js, Go or CLI client.

  • Store facts about users or entities with client.retain and search them later with client.recall
  • Ask a memory bank a question and get a disposition-aware answer through client.reflect
  • Add memory to an LLM app using the LLM Wrapper described in the README
  • Connect coding assistants through the MCP server or the documented coding agent integration

Great for

  • Developers building AI agents that need memory across sessions
  • Teams who want a self-hosted memory server with a choice of LLM providers
  • People who want to try a memory layer quickly through Docker
  • Developers who prefer a hosted option via Hindsight Cloud instead of running infrastructure

Not ideal for

  • People who only need a simple chat history buffer, since Hindsight is a larger system than that
  • Users who need a no-code app; the README is written for developers using clients and APIs
  • Anyone needing verified benchmark claims, since the README says other scores are self-reported by vendors
Step-by-step guide: how to use HindsightInstall, first run, examples, tips and FAQs

How to get it

  1. 1

    Run with Docker

    You'll need: Docker · An LLM API key or a local model provider such as Ollama

    Start server with Docker

    docker run -it --pull always --name hindsight --restart unless-stopped -p 8888:8888 -p 9999:9999 -e HINDSIGHT_API_LLM_API_KEY=$OPENAI_API_KEY -v hindsight-data:/home/hindsight/.pg0 ghcr.io/vectorize-io/hindsight:latest

    Install Python client

    pip install hindsight-client -U
  2. 2

    Run it

    • Set OPENAI_API_KEY in your shell, or choose another provider with HINDSIGHT_API_LLM_PROVIDER
    • Run the Docker command to start the server
    • Open the API at http://localhost:8888 and the UI at http://localhost:9999
    • Install the Python client and call client.retain, then client.recall against bank my-bank
  3. 3

    Get the most out of it

    • Use HINDSIGHT_API_LLM_PROVIDER to switch between hosted, local, or gateway LLM providers
    • Use the docker compose setup with external PostgreSQL for a more production-style deployment
    • Use the Helm chart for Kubernetes deployments
    • Read the supported models and benchmarks pages on the documentation site for provider-specific details

Downloads

Source code (.zip)main branch · straight from GitHubRelease v1.1.08 files incl. hindsight_all-1.1.0.tar.gzOfficial website & docshindsight.vectorize.io/

Strengths

  • Supports many LLM providers, including fully local options such as Ollama and LM Studio
  • Multiple install paths: Docker, pip, Helm on Kubernetes, or a managed cloud service
  • Released under the MIT license with official clients for Python, Node.js, Go and a CLI

Trade-offs

  • Benchmark claims are mostly vendor-reported according to the README, so results should be checked independently
  • Intel Macs need a special slim package and are only partially supported by the pip path
  • Running it yourself means managing a server and database, which adds setup work

How it compares

ProjectStarsTypeLevelUpdatedBest for
hindsight (this)57.9kAI model / agentintermediate4 days agoDevelopers building AI agents that need memory across sessions
ruflo80.6kDev toolintermediate4 days agoDevelopers already using Claude Code or Codex who want multiple agents working together
supermemory31.5kAppintermediate4 days agoDevelopers building AI agents or chat apps that need per-user memory
agent-framework14.5kLibraryintermediate4 days agoDevelopers building agents or multi-agent workflows they plan to run in production
PocketFlow11.3kLibraryintermediate3 months agoDevelopers who want a small, readable LLM framework they can inspect in full
PageIndex40kLibraryintermediate7 days agoDevelopers building question-answering over long PDFs such as annual reports or manuals

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.