Your AI has amnesia. Fix it.
What if your AI actually got smarter every session instead of starting over?
curl -sSL greymatter.axiomworks.ai/install | bash
The Problem
Your architecture decisions, your coding patterns, your preferences — gone. Every session, you re-explain. Every session, your AI starts as a stranger.
Context windows aren't memory. They're a junk drawer that empties when you close the tab. GreyMatter fixes this.
Architecture
GreyMatter isn't a chatbot. It's the infrastructure that turns AI coding agents into an autonomous development platform.
You already use AI to write code. But every session starts from scratch — no memory of what worked, what broke, or how your project is structured.
GreyMatter wraps around your existing AI agent. When it starts, GreyMatter injects your project's knowledge: past bugs, architecture decisions, coding patterns, and active work items. Your AI starts informed, not empty.
Works with: Claude Code, OpenAI Codex CLI, Google Gemini CLI — any tool that supports the MCP protocol.
Click to collapse ↑
The coordinator is the brain behind the operation. It decides what gets built, dispatches work to agents, and verifies quality before code ships.
Work dispatch: Submit a task, and the coordinator assigns it to an available agent. Run up to 8 agents in parallel.
Quality gates: Every completion runs through a tiered eval pipeline — compile, test, lint (Tier 1), then structural analysis (Tier 2). Bad code gets rejected automatically.
Auto-push: When code passes all gates, it's pushed to GitHub automatically. No commits sitting on disk.
Click to collapse ↑
MCP (Model Context Protocol) is how AI tools communicate. GreyMatter's plugin gives your AI 20+ tools it can call as naturally as reading a file or running a command.
Memory tools: gm_knowledge — search, create, and manage knowledge entries. gm_observe — capture observations from sessions.
Work tools: gm_work — create and manage work items. gm_dispatch — spawn agents to build in parallel.
Quality tools: gm_standards — enforce team rules. gm_security — scan for PII, secrets, and vulnerabilities.
Click to collapse ↑
Everything GreyMatter knows lives in a single SQLite database on your machine. Nothing leaves your computer. No cloud. No telemetry.
What it stores: Observations, patterns, corrections, expertise, quality standards, work items, and session history. Over 10,000 entries in a production system.
Semantic search: Ollama runs nomic-embed-text locally. Ask "how do we handle auth?" and find entries about JWT, OAuth, session cookies, and SAML — even if those words don't appear in your query.
Encryption at rest: Optional SQLCipher encryption with FIPS 140-3 compatible key derivation. Your knowledge, protected.
Click to collapse ↑
Cognitive Memory
GreyMatter models memory the way your brain does. Important knowledge gets reinforced. Stale knowledge decays naturally.
Your brain doesn't remember everything equally. Important memories get reinforced; unimportant ones fade. GreyMatter works the same way.
Every knowledge entry has a stability score that increases when it's used and decreases over time. The algorithm (FSRS-6) is the same one used by Anki, the most popular flashcard app in the world — proven by millions of users since 1885.
In practice: A bug fix you applied yesterday has high retention. An architecture decision from 3 weeks ago starts fading — unless your AI uses it, which resets the clock. Knowledge that matters stays; knowledge that doesn't, gracefully decays.
Click to collapse ↑
Not all knowledge is equal. A "this crashed the server" correction is more urgent than a "we prefer tabs over spaces" pattern. GreyMatter categorizes automatically:
Observations — raw data captured from sessions. High volume, fast decay. The raw material.
Patterns — recurring behaviors detected across sessions. "This team always writes tests first." Medium decay.
Expertise — deep technical knowledge. "FIPS 140-3 requires NIST-approved algorithms." Slow decay.
Corrections — mistakes and fixes. "Never use shell=True." Fast decay (bugs get fixed).
Gotchas — traps the team has hit. "macOS kills unsigned binaries with SIGKILL." Medium decay.
Click to collapse ↑
Traditional search finds exact words. Semantic search finds meaning. GreyMatter does both, and both run locally on your machine.
Full-text search (FTS5): Instant keyword matching across all knowledge. Type "database" and find every entry mentioning databases.
Semantic search (Ollama): Ask a question in plain English. "How do we handle user authentication?" finds entries about JWT tokens, OAuth flows, session cookies, and SAML — even if those exact words aren't in your query. Powered by nomic-embed-text running locally via Ollama.
Nothing leaves your machine. No cloud API calls. No data sent to embedding services. Your knowledge stays yours.
Click to collapse ↑
When you sleep, your brain consolidates memories — strengthening important ones, discarding noise, and resolving conflicts between competing memories. GreyMatter does this after every AI session.
Contradiction detection: If a new entry says "coordinator runs on Mac Mini" but an old entry says "coordinator runs on agenticai2," the older entry gets flagged as stale. No more acting on outdated information.
Observation promotion: When the same observation appears across multiple sessions, it gets promoted from a fleeting observation to a confirmed pattern, and eventually to permanent expertise.
The result: Your AI's knowledge self-curates. Important things get stronger. Outdated things fade. You never have to manually clean up your knowledge base.
Click to collapse ↑
The Status Line
GreyMatter adds a status line to every Claude Code session. At a glance: what your AI knows, what it's working on, and whether it's safe. Tell Claude “explain the status line” and it will walk you through it.
The status line appears in every Claude Code session with GreyMatter installed.
Features
Security
Every line of code your AI writes passes through an LLM. What else passes through? Your API keys. Your database credentials. Your customers' PII.
SecureLLM is also available as a standalone product. Learn more →
Expert Agents
An agent without context is just a fancy autocomplete. GreyMatter agents carry "souls" — persistent identity documents with domain expertise, coding standards, common patterns, and failure modes.
When GreyMatter dispatches a work item, it matches the task to the right expert. A Rust issue gets the Rust expert. A React component gets the TypeScript expert. The right brain for every job.
Autonomous work queue: Submit tasks, set priorities, walk away. Agents execute in parallel, run quality gates, and generate follow-up work items when something fails.
Create your own experts by dropping a markdown file in ~/.greymatter/soul/experts/
Editions
Developer Platform
GreyMatter isn't just a tool — it's a platform. Three transports, one API:
MCP — 20 tools that integrate directly with Claude Code, Codex CLI, and Gemini CLI. Memory, work queue, security, agents — all through natural conversation.
REST API — The Go coordinator exposes a full REST API. Access GreyMatter from any web app, mobile app, or dashboard. Every endpoint secured by API key + rate limiting.
Python SDK — Import GreyMatter directly into your Python 3.12+ applications. Six namespaces: gm.knowledge, gm.work, gm.agents, gm.security, gm.sync, gm.config.
Quick Start
1. Install GreyMatter
curl -sSL greymatter.axiomworks.ai/install | bashChecks for Claude Code, downloads GreyMatter, sets up your database, configures the MCP plugin. Two minutes.
2. Log into Claude Code
With an API key: export ANTHROPIC_API_KEY=your-key — or with Claude Pro: /login
3. Open your project and start Claude
cd your-project && claude — 20 tools loaded automatically. Type /mcp to see them.
4. Optional: Local LLM for semantic search
Ollama enables vector search. GreyMatter works without it (keyword search only).
curl -fsSL https://ollama.com/install.sh | sh && ollama pull nomic-embed-text