v0.1.0 — First Public Release

memor-ai v0.1.0

Measured memory for coding agents. Fire and forget — no API keys needed.

Install

pipx install memor-cli
memor install-hook
memor daemon

Highlights

• Zero API keys — local embeddings via model2vec (potion-base-8M, 256-dim, ~60MB)
• Sub-15ms recall — hybrid scoring (similarity + recency + kind weight + quality)
• Auto-ingest daemon — watches ~/.claude/projects/ for new transcripts
• Extractive distillation — TF-IDF + clustering, auto-classifies memories as decision/bugfix/lesson/snippet
• Feedback loop — tracks whether recalled memories actually get used, adjusts quality scores
• Secret redaction — API keys, tokens, connection strings redacted at ingest before storage
• Contradiction handling — newer memories with replacement cues auto-supersede stale ones
• Memory compaction — near-duplicates merged, unused memories auto-deactivated after 30 days
• Web dashboard — memor dashboard at localhost:8420 with efficiency metrics
• 153 tests across 33 test files
• Session-aware recall — won't re-inject context from the current session

Commands

memor install-hook    Install Claude Code hook + download model
memor daemon          Auto-ingest + distill background watcher
memor dashboard       Web dashboard on localhost:8420
memor query <text>    Search memories from the CLI
memor scan            Audit DB for leaked secrets
memor scan --purge    Redact secrets in place
memor reingest        Wipe DB and re-ingest everything
memor forget-stale    Deactivate memories unused for 30+ days
memor help            Full manual

Links

• PyPI: https://pypi.org/project/memor-cli/
• Docs: https://github.com/bnimit/memor-ai#readme