Claude Self Reflect

Claude Self-Reflect

Claude forgets everything. This fixes that.

Single 44MB binary. No databases. No containers. No API keys required.

Install | How It Works | MCP Tools | Paper | FAQ
v9.4 — Multi-Source Memory
CSR now remembers more than transcripts: task outcomes, plan documents, and a cross-project session registry — each absorbed at the lifecycle stage it belongs to.
Episodes carry real task state again (Claude Code's TodoWrite→TaskCreate rename had silently emptied them — found, fixed, and guarded with schema-miss telemetry).
Sub-millisecond search, ~150ms cached startup, 720+ tests, zero external dependencies.
Release notes

Table of Contents

- The Problem — Why Claude needs memory
- The Architecture — How CSR solves it
- The Pipeline — Progressive enrichment (9.3x improvement)
- The Paper — The research behind the retrieval design (preprint)
- Install — One-command install, consent-first activation
- What You'll Ask — Natural language, no syntax
- Performance | MCP Tools | Hooks | CLI
- AI Narratives | Upgrading | Troubleshooting

The Forgetting Problem

Claude starts fresh every session. Solutions you found, architectures you designed, bugs you debugged — all gone.

Context retention drops below 20% after 10 sessions. CSR fixes this with a single binary that gives Claude perfect memory.

No special syntax. No commands. Install once, and past context appears automatically when you need it.
Explore the full documentation →

One Binary. 44MB.

Everything runs locally in a single process. No Docker, no database server, no API keys required.

- SQLite — storage for chunks, embeddings, enrichment state
- HNSW — sub-millisecond vector search (
Explore the full documentation →

The Pipeline

Three layers progressively improve search quality from raw chunks to…