vectorize-io/hindsight
Hindsight: Agent Memory That Learns observed · 2026-08-28
Health v2 · maintenance only
79/100
- Activity 99
- Release rhythm 87
- Longevity 21
How is this computed?
round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-09-03. Adoption (stars, forks) is never an input.
- gap_med: 2
- age_days: 307
- days_rel: 8
- days_push: 7
- n_releases_24m: 68
Adoption not part of the score
21228 stars · 1638 forks observed · 2026-08-28
What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded
Hindsight is an open-source agent memory system that gives AI agents persistent long-term memory that learns and consolidates knowledge over time, using facts, observations, mental models, and graph-based relationships instead of raw RAG chunks. It runs as a self-hostable API server (Python, backed by PostgreSQL) with clients in Python, TypeScript, Go, and CLI, plus an MCP server and dozens of integrations with agent frameworks and coding agents.
Use cases
- add persistent long-term memory to my AI agent
- make my chatbot remember users between sessions
- give Claude Code or Cursor memory of my project's decisions
- replace RAG with a memory system that does temporal reasoning
- store and recall user preferences across conversations
- add memory to LangGraph or Vercel AI SDK agents
- self-host an agent memory server with MCP support
When to choose
- you need agents that remember and consolidate knowledge across sessions, not just retrieve documents
- you want state-of-the-art long-term memory accuracy (LongMemEval) with temporal and graph reasoning
- you need per-user or per-project memory banks with multi-tenant isolation
- you want a self-hostable, PostgreSQL-backed memory service with broad framework integrations
- you want coding agents to learn project conventions from git history and past sessions
When to avoid
- you only need simple vector search over static documents - plain RAG or a vector database is simpler
- you need a fully managed zero-ops solution and don't want to run a server or use Hindsight Cloud
- your agent is stateless by design and memory would add unwanted latency or complexity
- you require an LLM provider without structured output support, which Hindsight depends on
Facets
service · maturity active
rag vector-database agent-framework mcp llm-inference api-framework search-engine artificial-intelligence large-language-models developer-tools self-hosted python self-hosted cross-platform go cli agent-memory long-term-memory memory-banks temporal-reasoning knowledge-graph mental-models coding-agents langgraph vercel-ai-sdk mcp-server postgresql ai-agents retrieval-augmented-generation docker nodejs
10 sources
- readme: https://github.com/vectorize-io/hindsight · fetched 2026-08-28 · d40ec80a0225
- homepage: https://hindsight.vectorize.io/ · fetched 2026-08-29 · b9c039478d14
- site_page: https://hindsight.vectorize.io/developer/api/quickstart · fetched 2026-08-29 · 70a025fcc7ae
- site_page: https://hindsight.vectorize.io/integrations · fetched 2026-08-29 · 376424cd843f
- site_page: https://hindsight.vectorize.io/changelog · fetched 2026-08-29 · 2c315f8e32d5
- site_page: https://hindsight.vectorize.io/faq · fetched 2026-08-29 · 4bd74ef94ce1
- site_page: https://hindsight.vectorize.io/sdks/integrations/coding-agents · fetched 2026-08-29 · 556268647158
- site_page: https://hindsight.vectorize.io/sdks/integrations/langgraph · fetched 2026-08-29 · 6b99ba4574cb
- site_page: https://hindsight.vectorize.io/sdks/integrations/ai-sdk · fetched 2026-08-29 · 8c72061af7b2
- site_page: https://hindsight.vectorize.io/sdks/integrations/chat · fetched 2026-08-29 · 7994e2aca45f
Member repositories
| Repository | Role | Health v2 |
|---|---|---|
| vectorize-io/hindsight | main | 79 |
For agents
markdown · JSON · MCP: product_card(name="vectorize-io/hindsight")
Data as of 2026-08-30T08:39:29.467469+00:00 · Report a problem