# Hindsight

> Hindsight by Vectorize is an MIT-licensed memory server (v1.0.0, October 2026) with retain, recall and reflect operations, memory banks, an MCP endpoint per bank and plugins for coding agents such as Claude Code and Cursor. It stores memories in PostgreSQL with pgvector (embedded or external). It needs an LLM to extract facts: hosted APIs or local Ollama, LM Studio or llama.cpp. Runs via Docker, pip or Helm; a paid Hindsight Cloud also exists. Hardware requirements are not documented.

- Repository: https://github.com/vectorize-io/hindsight
- Official docs: https://hindsight.vectorize.io
- Checked: 2026-10-15

## What it replaces
- Mem0 Platform: Long-term memory for chat assistants, Memory for coding agents, Knowledge graph memory (partial). Covers the same memory jobs when self-hosted; you run the server and supply an LLM yourself.

## Facts, with sources

| Question | Status | Answer | Source | Checked |
|---|---|---|---|---|
| Runs on my machine? | Confirmed | Yes, Docker, pip or embedded Python on your own machine | https://github.com/vectorize-io/hindsight#quick-start | 2026-10-15 |
| Code licence | Confirmed | MIT | https://github.com/vectorize-io/hindsight/blob/main/LICENSE | 2026-10-15 |
| Model weights licence | Confirmed | No bundled LLM; licence depends on the model you connect | https://github.com/vectorize-io/hindsight#quick-start | 2026-10-15 |
| Commercial use | Confirmed | Yes, MIT code | https://github.com/vectorize-io/hindsight/blob/main/LICENSE | 2026-10-15 |
| Paid services needed? | Partial | Needs an LLM; local Ollama, LM Studio or llama.cpp work, hosted APIs and Hindsight Cloud are optional | https://github.com/vectorize-io/hindsight#1-start-a-server | 2026-10-15 |
| Hardware | Unknown | Not documented in the README; depends on the LLM you connect | https://github.com/vectorize-io/hindsight#readme | 2026-10-15 |
| Operating systems | Confirmed | Linux (x86\_64, ARM64), macOS (Apple Silicon; Intel via slim package) and Windows x86\_64 | https://github.com/vectorize-io/hindsight#supported-platforms | 2026-10-15 |
| Setup effort | Confirmed | Easy; one Docker command or pip install hindsight-all, plus an LLM provider setting | https://github.com/vectorize-io/hindsight#quick-start | 2026-10-15 |

GitHub: 50199 stars, 6092 forks, last push 2026-10-16, archived: no.
---
Source: RepoQuarry (https://repoquarry.dev/projects/hindsight). Data licence: CC BY 4.0. Quotes are from project documentation and belong to their authors.