vektor-slipstream

Hardware-accelerated persistent memory for AI agents. Local-first. No cloud. One-time payment.

66.9% on LoCoMo benchmark (adjusted). Under 1ms retrieval. Zero cloud dependency.

---

Install

npm install vektor-slipstream
npx vektor setup

Quick Start

const { createMemory } = require('vektor-slipstream');

const memory = await createMemory({
  agentId:    'my-agent',
  licenceKey: process.env.VEKTOR_LICENCE_KEY,
});

// Store a memory
await memory.remember('User prefers TypeScript over JavaScript');

// Recall by semantic similarity — sub-1ms, fully local
const results = await memory.recall('coding preferences', 5);
// → [{ content, score, id }]

// Traverse the MAGMA graph
const graph = await memory.graph('TypeScript', { hops: 2 });

// What changed in 7 days?
const delta = await memory.delta('project decisions', 7);

// Morning briefing
const brief = await memory.briefing();

// Graph stats
const stats = memory.graphStats();
// → { nodes, edges, entities }

---

What's New in v1.5.0

Retrieval pipeline rebuilt from scratch.

• bge-small-en-v1.5 bi-encoder + ms-marco cross-encoder reranker (spec-decode architecture)
• BM25 + Porter-stemmed BM25 + named entity injection, fused via RRF
• MAGMA graph layer — co-occurrence and temporal edges between entities in SQLite
• Persistent entity index (vektor_entities) for guaranteed named-entity recall
• Foresight extraction — future-tense statements stored for temporal queries
• Question type classifier — routes single-hop vs multi-hop to optimal retrieval path
• ADD-only contradiction detection — conflicting facts survive with timestamps (no silent deletes)
• Agentic sufficiency check — reformulates query if key entities missing from top results

LoCoMo benchmark results (conv 0, 154 valid questions):

Category	Judge Accuracy
Multi-hop	79.1%
Adversarial	70.4%
Temporal	46.2%
Single-hop	51.6%
Adjusted total	66.9%

#Under 1ms retrieval latency with zero cloud API calls at query time.

---

CLI Chat — Persistent Memory Terminal

Chat with any LLM with full memory across every session. Zero configuration.

npx vektor chat                          # start chat (auto-detects Ollama)
npx vektor chat --provider claude        # use Anthropic Claude
npx vektor chat --provider groq --model llama-3.3-70b-versatile
npx vektor chat --provider gemini
npx vektor chat --provider openai

Providers

Provider	Details
ollama	Default — free, local, no API key. Auto-detects best installed model.
claude	Anthropic Claude — set ANTHROPIC_API_KEY
openai	OpenAI GPT — set OPENAI_API_KEY
groq	Groq LLaMA — set GROQ_API_KEY (free tier available)
gemini	Google Gemini — set GEMINI_API_KEY

Set a permanent default:

# Windows
$env:VEKTOR_PROVIDER = "claude"

# macOS/Linux
export VEKTOR_PROVIDER=claude

In-chat commands

Type / to see available commands with autocomplete. Tab to select, arrow keys to navigate.

Command	Action
/recall	Search MAGMA memory mid-conversation
/stats	Show memory node count, edges, pinned
/briefing	Generate memory briefing inline
/exit	Exit chat (Ctrl+C also works)

One-liner commands

# Store a fact
npx vektor remember "I prefer TypeScript over JavaScript"
npx vektor remember "deadline is Friday" --importance 5

# Pipe support
cat meeting-notes.txt | npx vektor remember

# One-shot recall + LLM answer
npx vektor ask "what stack am I using?"
npx vektor ask "what did we decide about the database?"

# Autonomous goal executor
npx vektor agent "summarise everything I know about project Alpha"
npx vektor agent "research AI memory tools" --steps 15 --provider groq

Ollama auto-detection

VEKTOR queries http://localhost:11434/api/tags and picks the best available model: qwen3 → qwen2 → llama → mistral → first available.

Override:

$env:OLLAMA_MODEL = "qwen3.5:4b"
export OLLAMA_MODEL=qwen3.5:4b

---

All CLI Commands

npx vektor setup      # First-run wizard — licence, hardware, integrations
npx vektor activate   # Activate licence key on this machine
npx vektor test       # Test memory engine with progress bar
npx vektor status     # System health check
npx vektor mcp        # Start Claude Desktop MCP server
npx vektor rem        # Run REM dream cycle
npx vektor chat       # Persistent memory chat (all LLMs)
npx vektor remember   # Store a fact
npx vektor ask        # Query memory + LLM answer
npx vektor agent      # Autonomous goal executor
npx vektor help       # All commands

---

Claude Desktop Extension (DXT)

Install the .dxt extension for zero-config memory in every Claude Desktop session.

Install: drag vektor-slipstream.dxt onto the Claude Desktop Extensions page.

Once installed, Claude automatically:

• Recalls relevant context at session start
• Stores facts and decisions during conversation
• Summarises at session end

All 44 tools are available in Claude Desktop — no configuration needed beyond your licence key.

User config fields:

Field	Purpose
licence_key	Your Polar licence key (required)
db_path	Memory DB path (defaults to ~/vektor-slipstream-memory.db)
project_path	Default path for cloak_cortex project scanning (optional)

Download the latest .dxt from vektormemory.com/docs/dxt.

---

MCP Tools — All 44

Memory Tools

Tool	Function
vektor_recall	Semantic + BM25 + graph search across MAGMA memory
vektor_recall_rrf	BM25+RRF dual-channel recall with cross-encoder rerank
vektor_store	Store memory with importance score
vektor_ingest	Batch ingest conversation turns with session date
vektor_graph	Traverse associative memory graph
vektor_delta	See what changed on a topic over time
vektor_briefing	Generate morning briefing from recent memories
vektor_stats	Memory DB stats — node count, edges, entities
vektor_graph_stats	MAGMA graph node/edge/entity counts
vektor_timeline	Query memories by date range

CLOAK Core

Tool	Function
cloak_fetch	Stealth headless browser fetch via Playwright
cloak_fetch_smart	Checks llms.txt first, falls back to stealth browser
cloak_render	Full CSS/DOM layout sensor
cloak_diff	Semantic diff of URL since last fetch
cloak_diff_text	Structural diff between two text blobs
cloak_passport	AES-256-GCM credential vault (get/set/delete/list)
cloak_ssh_exec	Execute commands on remote server via SSH
cloak_ssh_upload	Upload file to remote server via SFTP
tokens_saved	Token efficiency ROI calculator

Identity Tools

Tool	Function
cloak_identity_create	Create persistent browser fingerprint identity
cloak_identity_use	Apply saved identity to a fetch call
cloak_identity_list	List all saved identities with trust summary

Behaviour Tools

Tool	Function
cloak_inject_behaviour	Human mouse/scroll injection for reCAPTCHA/Cloudflare bypass
cloak_behaviour_stats	List available patterns and categories
cloak_load_pattern	Load custom recorded behaviour pattern
cloak_pattern_stats	Self-improving pattern store tier breakdown
cloak_pattern_list	List patterns with scores and tier
cloak_pattern_prune	Remove stale/low-scoring patterns
cloak_pattern_seed	Seed store with built-in patterns

CAPTCHA Tools

Tool	Function
cloak_detect_captcha	Detect CAPTCHA type and sitekey
cloak_solve_captcha	Solve via vision AI (Claude/GPT-4o/2captcha)

Compression and Cortex Tools

Tool	Function
turbo_quant_compress	PolarQuant vector compression (~75% smaller)
turbo_quant_stats	Compression ratio and savings stats
cloak_cortex	Scan project directory into MAGMA entity graph
cloak_cortex_anatomy	Get cached file anatomy without rescanning

Multimodal Tools

Tool	Function
vektor_text	Text generation across providers (OpenAI/Claude/Groq/Gemini/NVIDIA NIM)
vektor_image	Image generation (DALL-E, Stability, NVIDIA)
vektor_vision	Image understanding and analysis
vektor_speech	Text-to-speech and transcription
vektor_search	Web search with memory integration
vektor_providers	List available multimodal providers and status

Agent Tools

Tool	Function
vektor_agent_run	Run autonomous goal executor with memory
vektor_swarm	Launch multi-agent swarm task
vektor_watch	File system watcher — auto-ingest on change

---

Claude Code Setup

Add to .claude/settings.json in your project:

{
  "mcpServers": {
    "vektor": {
      "command": "node",
      "args": ["/path/to/node_modules/vektor-slipstream/index.js"],
      "env": {
        "VEKTOR_LICENCE_KEY": "your-licence-key",
        "CLOAK_PROJECT_PATH": "/path/to/your/project"
      }
    }
  }
}

All 44 tools are available in Claude Code via this config.

---

What's Included

Memory Core (MAGMA)

• 4-layer associative graph — semantic, causal, temporal, entity
• MAGMA graph bridge — co-occurrence and temporal edges in SQLite (vektor-magma-bridge.js)
• bge-small-en-v1.5 bi-encoder + ms-marco cross-encoder reranker (vektor-embedder.js)
• BM25 + stemmed BM25 + RRF fusion — keyword + semantic dual-channel recall
• Persistent entity index — guaranteed named-entity retrieval
• Foresight extraction — future-tense statements stored with temporal metadata
• ADD-only contradiction detection — full history preserved, no silent overwrites
• AUDN curation loop — zero contradictions, zero duplicates
• REM dream cycle — up to 50:1 compression
• Sub-1ms recall — local SQLite, no network required
• Local ONNX embeddings — $0 embedding cost, no API key required

Integrations

• Claude Desktop — DXT extension, 44 tools, auto-memory system prompt
• Claude Code — MCP server, all 44 tools
• CLI — chat, remember, ask, agent commands
• LangChain — v1 + v2 adapter included
• OpenAI Agents SDK — drop-in integration
• Gemini · Groq · Ollama · NVIDIA NIM — provider agnostic

---

Performance

Metric	Value
Recall latency	sub-1ms (local SQLite + ONNX)
Embedding cost	$0 — fully local ONNX
Embedding latency	~10ms GPU / ~25ms CPU
LoCoMo benchmark	66.9% adjusted judge accuracy
vs Mem0	beats Mem0 old algorithm (62.47%)
First run	~2 min (downloads ~25MB model once)
Subsequent boots	<100ms

Hardware Auto-Detection

Zero config. VEKTOR detects and uses the best available accelerator:

• NVIDIA CUDA — GPU acceleration
• Apple Silicon — CoreML
• CPU — optimised fallback, works everywhere

---

Environment Variables

Variable	Default	Purpose
VEKTOR_SUMMARIZE	false	Enable LLM session summarization on ingest
VEKTOR_TRIPLES	true	Enable batch triple extraction on ingest
VEKTOR_FORESIGHT	true	Extract future-tense foresight signals
VEKTOR_TEMPORAL	true	Enable temporal index and date boosting
VEKTOR_CONTRADICT	true	Enable ADD-only contradiction detection
VEKTOR_DEBUG	—	Enable verbose retrieval debug output
VEKTOR_MODEL	Xenova/bge-small-en-v1.5	Swap embedding model (e.g. bge-large for higher accuracy)
VEKTOR_RERANK	true	Enable cross-encoder reranking

---

Licence

Commercial licence granted. Monthly fee - all updates included

Solo $9/mo → 3 licences | Team $35/mo → 5 licences | Studio $59/mo → 10 licences | Enterprise $99/mo → 25 licences |

Purchase: vektormemory.com/product#pricing Docs: vektormemory.com/docs Support: hello@vektormemory.com

---

Research

Built on peer-reviewed research:

• MAGMA (arxiv:2601.03236) — Multi-Graph Agentic Memory Architecture
• EverMemOS (arxiv:2601.02163) — Self-Organizing Memory OS
• HippoRAG (arxiv:2405.14831) — Neurobiologically Inspired Long-Term Memory (NeurIPS 2024)
• Mem0 (arxiv:2504.19413) — Production-Ready Agent Memory
• LoCoMo Benchmark — Long-Context Conversational Memory evaluation