MCPOfficialVectorize
Hindsight
Agent memory that learns from experience: it retains, recalls and reflects.
Vectorize’s Hindsight server stores facts, the agent’s experiences and the observations drawn from them in memory banks, and every bank has a built-in MCP endpoint with the retain, recall and reflect tools. It runs in Docker and needs a language model: an API key, a local model (such as Ollama) or an existing subscription; a separate package wires project memory into Claude Code, Codex and Cursor.
Installation
Manual install
$docker run -it --pull always --name hindsight --restart unless-stopped -p 8888:8888 -p 9999:9999 -e HINDSIGHT_API_LLM_API_KEY=<API_KEY> -v hindsight-data:/home/hindsight/.pg0 ghcr.io/vectorize-io/hindsight:latestDetails
- Works in
- Claude Code, Claude, Cursor, Codex, Gemini CLI, VS Code, Windsurf, Zed
- Transport
- local
- License
- MIT
- Tasks
- Research, Code
- Rating
- Source
- github.com/vectorize-io/hindsight (opens in a new tab)
Tags
- memory
- learning
- context
- agents
A pack is your own set of skills and MCP servers. sciver loads it in your AI with one command.