v0.1.0 Starter

Your agent forgets.
Fix that.

Guided memory routing for AI coding agents. Three backends, one router, zero context loss. Every query hits the right memory store — automatically.

$49 — one-time purchase, no subscription
openclaw-memory
# Install + activate
$ ./install.sh --key=oc-starter-xxxxxxxxxxxx
[OK] License verified (tier: starter)
[OK] Files installed
 
# Initialize in your project
$ openclaw-memory init
Orphan branch created: openclaw/memory
QMD collection indexed: 142 files
 
# Query — routed automatically
$ openclaw-memory "find function parseAuthToken"
Router: exact_symbol → QMD (BM25)
Found: src/auth/token.ts:42
// components

Three backends. One router.

Each memory backend is optimized for a different kind of recall. The router picks the right one for every query.

⇌

Memory Router

Rule-based query classification with sequential fallback. Pattern-matches your query to the right backend — no LLM call needed.

🧠

Total Recall

Git-native memory. Stores decisions, context, and learned patterns as markdown on an orphan branch. Versioned. Portable. Searchable.

⚡

QMD

BM25 + vector search over your codebase. Instant symbol lookup, semantic concept search, and hybrid queries. Powered by Bun.

// routing logic

Queries find the right store.

The router classifies every query and picks the optimal backend with automatic fallback.

Symbol lookup
QMD (BM25) → Total Recall
Concept search
QMD (vector) → Total Recall
Recent context
Total Recall → QMD
File patterns
QMD (BM25) → Total Recall
Ambiguous
QMD (hybrid) → Total Recall
$49
One-time purchase. No subscription. No SaaS. Just files.