memor-cli 0.6.0

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

                                                _
 _ __ ___   ___ _ __ ___   ___  _ __       __ _(_)
| '_ ` _ \ / _ \ '_ ` _ \ / _ \| '__|____ / _` | |
| | | | | |  __/ | | | | | (_) | | |_____| (_| | |
|_| |_| |_|\___|_| |_| |_|\___/|_|        \__,_|_|

  Measured memory for coding agents.

Automatic background memory for Claude Code. Fire and forget — no API keys needed.

Memor watches your coding sessions, extracts decisions and patterns, and recalls relevant context on every prompt. Zero configuration. One install. Your agent remembers everything.

---

Quick Start

# Install globally (recommended)
pipx install memor-cli

# Install the Claude Code hook + download embedding model (~60MB)
memor install-hook

# Start as a background service (macOS/Linux)
memor service install

# Or run in the foreground
memor daemon

That's it. Every Claude Code prompt now gets automatic context recall. Open the dashboard to see it working:

memor dashboard
# Opens http://localhost:8420

Alternative install: pip install memor-cli works too — just make sure ~/.local/bin is on your PATH so the memor command is available.

---

How It Works

  You type a prompt in Claude Code
      |
      v
  Hook fires (UserPromptSubmit)
      |
      v
  Embed query locally (model2vec, ~2ms)
      |
      v
  Hybrid retrieval: dense vectors + lexical BM25, fused (RRF)
      |
      v
  Relevance gate drops off-topic matches (inject nothing if nothing fits)
      |
      v
  Rank: similarity + recency + kind weight + quality
      |
      v
  Inject relevant context into prompt
      |
      v
  Claude sees your past decisions, bugfixes,
  architecture choices — without you re-explaining

Two background processes:

1. Daemon — polls ~/.claude/projects/ for transcripts, embeds chunks, runs distillation, analyzes feedback (positive and negative), promotes cross-project patterns to global scope, compacts duplicates, tracks session-level token usage. All local.
2. Hook — fires on every prompt, recalls relevant memories, injects them as context. Sub-15ms.

No API keys required. Embeddings run locally via model2vec (potion-base-8M, 256-dim). Vectors stored in sqlite-vec. Everything runs on your machine.

---

Hybrid Retrieval

Memor retrieves over two channels and fuses them, so it catches both semantic matches and exact terms:

• Dense — local vector similarity (model2vec) for semantic recall.
• Lexical — SQLite FTS5 / BM25 over the raw text, to recover exact identifiers, error strings, and API names that static embeddings blur together.

The two rankings are combined with Reciprocal Rank Fusion (RRF). A relevance gate drops anti-correlated (off-topic) candidates before ranking, so an unrelated prompt injects nothing rather than the least-bad guess. The lexical channel only activates when the dense channel finds the query on-topic, preventing generic words from pulling in noise.

Tunable via MEMOR_MIN_SIMILARITY (relevance floor, default 0.0) and MEMOR_MAX_TOKENS (injection budget, default 1500).

Scoring

Surviving candidates are ranked by four signals:

Signal	Weight	How it works
Semantic similarity	50%	Dense + lexical relevance, fused via RRF
Recency	25%	Exponential decay with 14-day half-life — recent decisions rank higher
Kind weight	15%	Distilled memories (1.3x) rank above raw session chunks (1.0x)
Quality	10%	Bayesian score from implicit feedback — memories the agent actually uses rank higher

This means a relevant decision from yesterday beats a vaguely-related chunk from a month ago — even if the raw embedding similarity is similar.

Feedback Loop

Memor tracks whether recalled memories actually get used by the agent — and whether they actively hurt. After each session, the daemon analyzes the transcript in both directions:

• Positive signal — n-gram overlap or semantic similarity between recalled content and the agent's response. Memories that consistently prove useful get quality boosts.
• Negative signal — user rejection ("no that's wrong", "we switched to X") or assistant contradiction ("however, looking at the current code, we actually use Y"). Memories that get corrected receive a quality penalty, making them less likely to be recalled next time.

The quality formula is Bayesian: (uses - negatives + 1) / (recalls + 2). One correction weighs as much as one positive use, so harmful memories drop fast. Memories never recalled in 30+ days get automatically deactivated. Near-duplicate memories are compacted into one.

---

What Gets Stored

Kind	Source	Description
session_chunk	Daemon auto-ingest	Filtered turns from Claude Code transcripts
memory	Extractive distillation	Key decisions, patterns, bugfixes per session

Memories are automatically classified as decision, bugfix, lesson, snippet, or generic extract based on content patterns. The daemon runs a signal filter that keeps decisions, bugfixes, lessons, and code rationale while skipping noise (tool calls, file listings, boilerplate).

---

Global Memories

Some patterns aren't project-specific — they're yours. "Always use type hints." "Structure FastAPI apps with a routes/ directory." "Prefer composition over inheritance."

Memor detects these automatically. When the same pattern appears in 3 or more projects (measured by embedding similarity), the daemon promotes it to a _global scope:

• Global memories are recalled everywhere — they show up in every project's search results alongside project-specific memories.
• Source duplicates are deactivated — the per-project copies get superseded by the single global version, reducing clutter.
• No manual tagging — promotion is fully automatic, based on cross-project clustering.

This means your coding habits and preferences follow you into new projects from the first prompt, without you having to re-explain anything.

---

Dashboard

memor dashboard

Dark fintech-inspired UI showing:

• Hero metrics — total memories, recall count, avg latency, coverage — with sparkline bars
• Daily recall activity — stacked bar chart of hits vs misses over time
• Session efficiency — real token savings measured from API usage data (avg tokens/turn with vs without recall)
• Per-project breakdown — artifact counts, token totals, last activity
• Recent recalls — every hook event with scores, latency, and status

---

Commands

memor help                           Print the full manual
memor install-hook                   Install Claude Code hook + download model
memor daemon                         Auto-ingest + distill (background watcher)
memor dashboard                      Web dashboard on localhost:8420
memor version                        Print installed version
memor service install                Run daemon as background service (launchd/systemd)
memor service stop                   Stop the background service
memor service uninstall              Remove the background service
memor service status                 Check if the service is running
memor query <text>                   Search memories from the CLI
memor reingest                       Wipe DB and re-ingest everything
memor reingest --project <name>      Re-ingest only one project
memor forget-stale                   Deactivate memories unused for 30+ days
memor scan                           Audit DB for leaked secrets
memor scan --purge                   Redact secrets in place
memor setup-model                    Download/retry the embedding model
memor ingest-cc <file>               Ingest a single transcript
memor ingest-project <dir>           Bulk ingest a project directory
memor ingest-doc <file>              Ingest a markdown document
memor distill --project <name>       Run distillation manually
memor eval <cases.json>              Run eval suite
memor eval-counterfactual --project  Win/tie/loss vs no-memory baseline
memor bench-embed --project <name>   Compare embedding models

---

Architecture

memor/
+-- types.py              Core dataclasses: Artifact, Scope, Hit, RetrievalTrace
+-- interfaces.py         Protocols: Embedder, LLM, MemoryStore
+-- cli.py                Typer CLI entry point
+-- daemon.py             Auto-ingest + auto-distill + compaction watcher
+-- project.py            Git-root project resolver (filesystem-aware)
+-- recall.py             Shared recall core (used by hook + skill)
+-- redact.py             Secret detection and redaction at ingest
+-- feedback.py           Feedback analyzer (positive usage + negative signals)
+-- global_memories.py    Cross-project promotion to _global scope
|
+-- retrieve/
|   +-- retriever.py      Hybrid retrieval (dense + BM25, RRF) + relevance gate + scoring
|
+-- store/
|   +-- sqlite_store.py   SQLite + sqlite-vec + FTS5 (WAL mode, dimension safety)
|
+-- embed/
|   +-- local.py          model2vec (potion-base-8M, 256-dim, ~60MB)
|   +-- api.py            OpenAI-compatible embedding API (optional)
|   +-- fake.py           Deterministic SHA-256 embedder (tests)
|
+-- service.py            Background service management (launchd/systemd)
+-- dashboard/
|   +-- server.py         FastAPI dashboard backend
|   +-- static/index.html Self-contained dashboard (no CDN deps)
|
+-- distill/
|   +-- extractive.py     TF-IDF + clustering + auto-classification
|   +-- distiller.py      Extractive + optional LLM abstractive
|
+-- eval/
    +-- runner.py          4-baseline eval runner
    +-- judge.py           LLM-as-judge evaluation
    +-- embed_benchmark.py Embedding model comparison

memor/hook_cli.py          Claude Code hook entry point (thin client)
skill/recall.py            Standalone recall script

---

Security

Nothing leaves your machine. In the default configuration:

• No telemetry, no analytics, no phone-home. Zero outbound network calls.
• Embeddings run locally via model2vec static token embeddings — no inference runtime, no GPU (one-time model download from HuggingFace — no user data sent).
• Hook transport is a Unix socket (~/.memor/hook.sock), not a network port.
• Dashboard binds localhost only.

The only optional network paths are the LLM-based abstractive distiller (requires explicitly setting ANTHROPIC_API_KEY) and the API embedding backend — both off by default.

Secret redaction

Memor automatically redacts secrets at ingest, before anything is embedded or stored:

• API keys (AWS AKIA..., OpenAI sk-..., Anthropic sk-ant-..., GitHub ghp_..., Stripe, Slack)
• JWTs, PEM private key blocks
• Connection strings (postgres://, mongodb://, redis://, etc.)
• .env-style assignments (DB_PASSWORD=..., API_KEY=...)
• High-entropy tokens (Shannon entropy > 4.0, length > 20)

Redacted content is replaced with [REDACTED] in place, preserving surrounding context. To audit and clean an existing database: memor scan (audit) or memor scan --purge (redact in place).

Contradiction handling

When a new memory contradicts an older one in the same project (detected via replacement cues like "switched from X to Y", "no longer", "ripped out"), the older memory is automatically deactivated. This prevents stale decisions from being recalled and misleading the agent.

Local storage

The memory database (~/.memor/memor.db) is stored as plaintext SQLite on disk. For at-rest protection, we recommend enabling OS-level full-disk encryption (FileVault on macOS, LUKS on Linux) which covers all local files with zero performance overhead.

---

Development

git clone https://github.com/bnimit/memor-ai.git
cd memor-ai
python3 -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"

pytest  # 251 tests

---

License

MIT. See LICENSE for the full text.