What's New in v2.1.2 "Honest Memory"

v2.1.2 makes Vestige easier to trust in everyday work: literal lookups stay literal, purge really removes content, contradictions are inspectable, and updates no longer require a curl reinstall flow.

• Concrete search mode. Quoted strings, env vars, UUIDs, paths, and code identifiers now take a keyword/literal path that skips HyDE, semantic fusion, FSRS reweighting, competition, and spreading activation. Exact things like OPENAI_API_KEY, mlx_lm.server, and migration IDs land first.
• Irreversible purge. memory(action="purge", confirm=true) permanently removes memory content and embeddings, scrubs insight JSON references, detaches temporal-summary children, prunes graph edges, and keeps only a non-content deletion tombstone for sync/audit.
• First-class contradiction inspection. New contradictions MCP tool surfaces trust-weighted disagreements directly instead of hiding them inside deep_reference.
• Simple update flow. vestige update and vestige sandwich install refresh binaries and companion files without making users paste curl commands.
• Pro waitlist preview. /dashboard/waitlist adds a local-first Solo Pro and Team Pro early-access surface. VITE_WAITLIST_ENDPOINT and VITE_SUPPORT_BOT_ENDPOINT are opt-in dashboard env vars, so no signup data is captured unless endpoints are configured.

What's New in v2.1.1 "Portable Sync"

v2.1.1 focuses on the biggest post-launch ask: move memories between machines without losing cognitive state. It also adds opt-in Qwen3 embeddings for higher-recall local retrieval.

• Exact portable archives. vestige portable-export / vestige portable-import preserve IDs, FSRS state, graph edges, suppression state, audit rows, and embedding blobs for Vestige-to-Vestige device transfer.
• Sync-safe merge storage. vestige portable-import --merge and vestige sync <archive> merge non-empty databases, apply delete tombstones, keep newer local memories, rebuild FTS, and push through a pluggable portable-sync backend. v2.1.1 ships the file backend for Dropbox, iCloud, Syncthing, Git, and shared folders.
• Qwen3 embeddings. Build with qwen3-embeddings, set VESTIGE_EMBEDDING_MODEL=qwen3-0.6b, and run vestige consolidate to re-embed existing memories. vestige health reports mixed-model stores before search quality is affected.
• Model-aware retrieval. Vestige now avoids comparing Qwen and Nomic vectors in the same search/dedup path.

What's New in v2.1.0 "Cognitive Sandwich Goes Local"

v2.1.0 adds an opt-in Claude Code hook harness around the existing Vestige MCP server. The MCP tool surface and database schema stay backward compatible, while preflight hooks can inject trusted memory context before Claude answers. The heavyweight Sanhedrin verifier is optional and can be enabled separately.

• Optional Sanhedrin Executioner. The post-response verifier is off by default. Users can enable it with an OpenAI-compatible endpoint on x86/Linux/Intel Mac, or add --with-launchd on Apple Silicon to run the local MLX Qwen backend.
• One-command Cognitive Sandwich installer. vestige sandwich install stages hook files and agents by default, removes old Vestige hook wiring, and leaves all Claude Code hook layers plus the 19 GB model path opt-in.
• Pulse hook backed by /api/changelog. Fresh dream and connection events can be injected into the next Claude Code prompt context without blocking the prompt.
• VESTIGE_DATA_DIR support. --data-dir now has an env-var fallback, tilde expansion, secure directory creation, and clear precedence docs.
• NPM release wrapper fixed. vestige-mcp-server@2.1.0 now downloads binaries from the matching v2.1.0 GitHub release tag instead of an old hardcoded release.

What's New in v2.0.9 "Autopilot"

Autopilot flips Vestige from passive memory library to self-managing cognitive surface. Same 24 MCP tools, zero schema changes — but the moment you upgrade, 14 previously dormant cognitive primitives start firing on live events without any tool call from your client.

• One supervised backend task subscribes to the 20-event WebSocket bus and routes six event classes into the cognitive engine: MemoryCreated triggers synaptic-tagging PRP + predictive-access records, SearchPerformed warms the speculative-retrieval model, MemoryPromoted fires activation spread, MemorySuppressed emits the Rac1 cascade wave, high-importance ImportanceScored (>0.85) auto-promotes, and Heartbeat rate-limit-fires find_duplicates on large DBs. The engine mutex is never held across .await, so MCP dispatch is never starved.
• Panic-resilient supervisors. Both background tasks run inside an outer supervisor loop — if one handler panics on a bad memory, the supervisor respawns it in 5 s instead of losing every future event.
• Fully backward compatible. No new MCP tools. No schema migration. Existing v2.0.8 databases open without a single step. Opt out with VESTIGE_AUTOPILOT_ENABLED=0 if you want the passive-library contract back.
• 3,091 LOC of orphan v1.0 tool code removed — nine superseded modules (checkpoint, codebase, consolidate, ingest, intentions, knowledge, recall, plus helpers) verified zero non-test callers before deletion. Tool surface unchanged.

What's New in v2.0.8 "Pulse"

v2.0.8 wires the dashboard through to the cognitive engine. Eight new surfaces expose the reasoning stack visually — every one was MCP-only before.

• Reasoning Theater (/reasoning) — Cmd+K Ask palette over the 8-stage deep_reference pipeline (hybrid retrieval → cross-encoder rerank → spreading activation → FSRS-6 trust → temporal supersession → contradiction analysis → relation assessment → template reasoning chain). Evidence cards, confidence meter, contradiction geodesic arcs, superseded-memory lineage, evolution timeline. Zero LLM calls, 100% local.
• Pulse InsightToast — real-time toasts for DreamCompleted, ConsolidationCompleted, ConnectionDiscovered, promote/demote/suppress/unsuppress, Rac1CascadeSwept. Rate-limited, auto-dismiss, click-to-dismiss.
• Memory Birth Ritual (Terrarium) — new memories materialize in the 3D graph on every MemoryCreated: elastic scale-in, quadratic Bezier flight path, glow sprite fade-in, Newton's Cradle docking recoil. 60-frame sequence, zero-alloc math.
• 7 more dashboard surfaces — /duplicates, /dreams, /schedule, /importance, /activation, /contradictions, /patterns. Left nav expanded 8 → 16 with single-key shortcuts.
• Intel Mac (x86_64-apple-darwin) support restored via the ort-dynamic Cargo feature + Homebrew onnxruntime. Microsoft deprecated x86_64 macOS prebuilts; the dynamic-link path sidesteps that permanently. Closes #41.
• Contradiction-detection false positives eliminated — four thresholds tightened so adjacent-domain memories no longer flag as conflicts. On an FSRS-6 query this collapses false contradictions 12 → 0 without regressing legitimate test cases.

What's New in v2.0.7 "Visible"

Hygiene release closing two UI gaps and finishing schema cleanup. No breaking changes, no user-data migrations.

• POST /api/memories/{id}/suppress + /unsuppress HTTP endpoints — dashboard can trigger Anderson 2025 SIF + Rac1 cascade without dropping to raw MCP. suppressionCount, retrievalPenalty, reversibleUntil, labileWindowHours all in response. Suppress button joins Promote / Demote / Delete on the Memories page.
• Uptime in the sidebar footer — the Heartbeat event has carried uptime_secs since v2.0.5 but was never rendered. Now shows as up 3d 4h / up 18m / up 47s.
• execute_export panic fix — unreachable match arm replaced with a clean "unsupported export format" error instead of unwinding through the MCP dispatcher.
• predict surfaces predict_degraded: true on lock poisoning instead of silently returning empty vecs. memory_changelog honors start / end bounds. intention check honors include_snoozed.
• Migration V11 — drops dead knowledge_edges + compressed_memories tables (added speculatively in V4, never used).

What's New in v2.0.6 "Composer"

v2.0.6 is a polish release that makes the existing cognitive stack finally feel alive in the dashboard and stays out of your way on the prompt side.

• Six live graph reactions, not one — MemorySuppressed, MemoryUnsuppressed, Rac1CascadeSwept, Connected, ConsolidationStarted, and ImportanceScored now light the 3D graph in real time. v2.0.5 shipped suppress but the graph was silent when you called it; consolidation and importance scoring have been silent since v2.0.0. No longer.
• Intentions page actually works — fixes a long-standing bug where every intention rendered as "normal priority" (type/schema drift between backend and frontend) and context/time triggers surfaced as raw JSON.
• Opt-in composition mandate — the new MCP instructions string stays minimal by default. Opt in to the full Composing / Never-composed / Recommendation composition protocol with VESTIGE_SYSTEM_PROMPT_MODE=full when you want it, and nothing is imposed on your sessions when you don't.

What's New in v2.0.5 "Intentional Amnesia"

The first shipped AI memory system with top-down inhibitory control over retrieval. Other systems implement passive decay — memories fade if you don't touch them. Vestige v2.0.5 also implements active suppression: the new suppress tool compounds a retrieval penalty on every call (up to 80%), a background Rac1 worker fades co-activated neighbors over 72 hours, and the whole thing is reversible within a 24-hour labile window. Never deletes. The memory is inhibited, not erased.

Ebbinghaus 1885 models what happens to memories you don't touch. Anderson 2025 models what happens when you actively want to stop thinking about one. Every other AI memory system implements the first. Vestige is the first to ship the second.

Based on Anderson et al. 2025 (Suppression-Induced Forgetting, Nat Rev Neurosci) and Cervantes-Sandoval et al. 2020 (Rac1 synaptic cascade).

Quick Start

# 1. Install
npm install -g vestige-mcp-server@latest

# 2. Connect to Claude Code
claude mcp add vestige vestige-mcp -s user

# Or connect to Codex
codex mcp add vestige -- vestige-mcp

# 3. Test it
# "Remember that I prefer TypeScript over JavaScript"
# ...new session...
# "What are my coding preferences?"
# → "You prefer TypeScript over JavaScript."

vestige update

vestige update --install-dir /usr/local/bin

brew install onnxruntime
npm install -g vestige-mcp-server@latest
echo 'export ORT_DYLIB_PATH="'"$(brew --prefix onnxruntime)"'/lib/libonnxruntime.dylib"' >> ~/.zshrc
source ~/.zshrc
claude mcp add vestige vestige-mcp -s user

npm install -g vestige-mcp-server@latest
vestige-mcp --version

{
  "mcpServers": {
    "vestige": {
      "command": "vestige-mcp"
    }
  }
}

git clone https://github.com/samvallad33/vestige && cd vestige
cargo build --release -p vestige-mcp

npm install -g vestige-mcp-server

git clone https://github.com/samvallad33/vestige && cd vestige
cargo build --release -p vestige-mcp
# Optional: enable Metal GPU acceleration on Apple Silicon
cargo build --release -p vestige-mcp --features metal

Works Everywhere

Vestige speaks MCP — the universal protocol for AI tools. One brain, every IDE.

🧠 3D Memory Dashboard

Vestige v2.0 ships with a real-time 3D visualization of your AI's memory. Every memory is a glowing node in 3D space. Watch connections form, memories pulse when accessed, and the entire graph come alive during dream consolidation.

Features:

• Force-directed 3D graph with 1000+ nodes at 60fps
• Bloom post-processing for cinematic neural network aesthetic
• Real-time WebSocket events: memories pulse on access, burst on creation, fade on decay
• Dream visualization: graph enters purple dream mode, replayed memories light up sequentially
• FSRS retention curves: see predicted memory decay at 1d, 7d, 30d
• Command palette (Cmd+K), keyboard shortcuts, responsive mobile layout
• Installable as PWA for quick access

Tech: SvelteKit 2 + Svelte 5 + Three.js + Tailwind CSS 4 + WebSocket

Run vestige dashboard to open http://localhost:3927/dashboard, or set VESTIGE_DASHBOARD_ENABLED=true to start it with the MCP server.

Architecture

┌─────────────────────────────────────────────────────┐
│  SvelteKit Dashboard (apps/dashboard)                │
│  Three.js 3D Graph · WebGL + Bloom · Real-time WS   │
├─────────────────────────────────────────────────────┤
│  Axum HTTP + WebSocket Server (port 3927)            │
│  15 REST endpoints · WS event broadcast              │
├─────────────────────────────────────────────────────┤
│  MCP Server (stdio JSON-RPC)                         │
│  25 tools · 30 cognitive modules                     │
├─────────────────────────────────────────────────────┤
│  Cognitive Engine                                    │
│  ┌──────────┐ ┌──────────┐ ┌───────────────┐       │
│  │ FSRS-6   │ │ Spreading│ │ Prediction    │       │
│  │ Scheduler│ │ Activation│ │ Error Gating  │       │
│  └──────────┘ └──────────┘ └───────────────┘       │
│  ┌──────────┐ ┌──────────┐ ┌───────────────┐       │
│  │ Memory   │ │ Synaptic │ │ Hippocampal   │       │
│  │ Dreamer  │ │ Tagging  │ │ Index         │       │
│  └──────────┘ └──────────┘ └───────────────┘       │
├─────────────────────────────────────────────────────┤
│  Storage Layer                                       │
│  SQLite + FTS5 · USearch HNSW · Nomic Embed v1.5    │
│  Optional: Nomic v2 MoE · Qwen3 Reranker · Metal   │
└─────────────────────────────────────────────────────┘

Why Not Just Use RAG?

RAG is a dumb bucket. Vestige is an active organ.

🔬 The Cognitive Science Stack

This isn't a key-value store with an embedding model bolted on. Vestige implements real neuroscience:

Prediction Error Gating — The hippocampal bouncer. When new information arrives, Vestige compares it against existing memories. Redundant? Merged. Contradictory? Superseded. Novel? Stored with high synaptic tag priority.

FSRS-6 Spaced Repetition — 21 parameters governing the mathematics of forgetting. Frequently-used memories stay strong. Unused memories naturally decay. Your context window stays clean.

HyDE Query Expansion (v2.0) — Template-based Hypothetical Document Embeddings. Expands queries into 3-5 semantic variants, embeds all variants, and searches with the centroid embedding for dramatically better recall on conceptual queries.

Synaptic Tagging — A memory that seemed trivial this morning can be retroactively tagged as critical tonight. Based on Frey & Morris, 1997.

Spreading Activation — Search for "auth bug" and find the related JWT library update from last week. Memories form a graph, not a flat list. Based on Collins & Loftus, 1975.

Dual-Strength Model — Every memory has storage strength (encoding quality) and retrieval strength (accessibility). A deeply stored memory can be temporarily hard to retrieve — just like real forgetting. Based on Bjork & Bjork, 1992.

Memory Dreaming — Like sleep consolidation. Replays recent memories to discover hidden connections, strengthen important patterns, and synthesize insights. Dream-discovered connections persist to a graph database. Based on the Active Dreaming Memory framework.

Waking SWR Tagging — Promoted memories get sharp-wave ripple tags for preferential replay during dream consolidation. 70/30 tagged-to-random ratio. Based on Buzsaki, 2015.

Autonomic Regulation — Self-regulating memory health. Auto-promotes frequently accessed memories. Auto-GCs low-retention memories. Consolidation triggers on 6h staleness or 2h active use.

Active Forgetting (v2.0.5) — Top-down inhibitory control via the suppress tool. Other memory systems implement passive decay — the Ebbinghaus 1885 "use it or lose it" curve, sometimes with trust-weighted strength factors. Vestige v2.0.5 also implements active top-down suppression: each suppress call compounds (Suppression-Induced Forgetting, Anderson 2025), a background Rac1 cascade worker fades co-activated neighbors across the connection graph (Cervantes-Sandoval & Davis 2020), and a 24-hour labile window allows reversal (Nader reconsolidation semantics on a pragmatic axis). The memory persists — it's inhibited, not erased. Explicitly distinct from Anderson 1994 retrieval-induced forgetting (bottom-up, passive competition during retrieval), which is a separate, older primitive that several other memory systems implement. Based on Anderson et al., 2025 and Cervantes-Sandoval et al., 2020. First shipped AI memory system with this primitive.

Full science documentation ->

🛠 25 MCP Tools

Context Packets

Core Memory

Cognitive Engine

Autonomic

Scoring & Dedup

Maintenance

Deep Reference (v2.0.4)

Active Forgetting (v2.0.5)

Make Your AI Use Vestige Automatically

Add this to your CLAUDE.md:

## Memory

At the start of every session:
1. Search Vestige for user preferences and project context
2. Save bug fixes, decisions, and patterns without being asked
3. Create reminders when the user mentions deadlines

Full CLAUDE.md templates ->

Technical Details

Optional Features

# Qwen3 embeddings (Candle backend; add metal on Apple Silicon)
cargo build --release -p vestige-mcp --features qwen3-embeddings,metal
VESTIGE_EMBEDDING_MODEL=qwen3-0.6b vestige consolidate

CLI

vestige stats                    # Memory statistics
vestige stats --tagging          # Retention distribution
vestige stats --states           # Cognitive state breakdown
vestige health                   # System health check
vestige consolidate              # Run memory maintenance
vestige restore <file>           # Restore from backup
vestige portable-export <file>         # Exact cross-device archive
vestige portable-import <file>         # Import archive into an empty database
vestige portable-import <file> --merge # Merge archive into this database
vestige sync <file>                    # Pull/merge/push via file backend
vestige dashboard                # Open 3D dashboard in browser

Documentation

Troubleshooting

which vestige-mcp
# Or use the full path:
claude mcp add vestige /usr/local/bin/vestige-mcp -s user

export HTTPS_PROXY=your-proxy:port

curl http://localhost:3927/api/health
# Should return {"status":"healthy",...}

More troubleshooting ->

Contributing

Issues and PRs welcome. See CONTRIBUTING.md.

License

AGPL-3.0 — free to use, modify, and self-host. If you offer Vestige as a network service, you must open-source your modifications.

Built by @samvallad33
_{80,000+ lines of Rust · 30 cognitive modules · 130 years of memory research · one 22MB binary}