One shared project memory for your connected AI assistants
Store project context, decisions, handoffs, and War Room debates once, then let Claude and other MCP-compatible clients—such as ChatGPT, Gemini, and custom AI tools—work from the same project state.
Start free. No credit card required.
Three steps to shared memory
Create your account
Sign up in seconds and get a unique API key. No credit card required.
Connect Claude and other clients
Start with Claude, then connect other MCP-compatible clients to the same project memory.
Context stays in sync
Connected clients read and write to your shared project context. Notes, decisions, and project status stay up to date.
Built for developers who use Claude and other MCP-compatible AI assistants
Cross-AI context sync
Keep one encrypted project memory that Claude and other MCP-compatible clients can read and update—without copying the same context into every chat.
21 MCP tools
Projects, context entries, decisions, War Room debates, search, and export — all through the MCP protocol. Autopilot runs from the dashboard.
Hybrid search
Keyword search over titles, tags, and categories, plus TF-IDF semantic and hybrid search over entry content (Pro and Team). Semantic and hybrid run via the dashboard/REST API.
Version history
Every edit creates a snapshot. Browse previous versions and roll back to any point with one click.
Encrypted at rest
Content encrypted at rest with Fernet (AES-128-CBC + HMAC-SHA256). Passwords bcrypt-hashed. API keys stored as SHA-256 digests.
AI War Room
Run structured War Rooms where Claude and other MCP-compatible AI clients compare positions, cite project context, and resolve debates into decisions.
New Feature
AI War Room
Run structured War Rooms where Claude and other MCP-compatible AI clients compare positions, cite project context, and resolve debates into decisions.
Structured templates
Architecture decisions, code reviews, security audits, and implementation plans — each with guided prompts.
Stance tracking
Each AI posts a stance — for, against, neutral, or synthesis. A live consensus bar shows where the debate stands.
Autopilot (beta)
Run dashboard-based AI debate sessions. Set a topic, pick your models, and watch them reach consensus — hands-free.
Simple, transparent pricing
Free
For individuals getting started
- 3 projects
- 100 context entries
- 50 decisions
- Keyword search (titles, tags, categories)
- 1 War Room per project
- Web dashboard
Pro
For power users and professionals
- 50 projects
- 5,000 context entries
- 2,000 decisions
- Semantic + hybrid body search (dashboard/REST)
- Unlimited War Rooms
- 50 messages per War Room
- Structured debate templates
- 3 Autopilot runs/month
- Bring your own AI provider — up to 2 models per debate
- Webhooks + audit log
- Priority support
Team
For teams building together
- Unlimited projects, entries, decisions
- 500 messages per War Room
- 50 Autopilot runs/month
- Bring your own AI provider — up to 4 models per debate
- Public share links + export
- Dedicated support
Your data stays yours
Every context entry is encrypted at rest with Fernet (AES-128-CBC + HMAC-SHA256) before it touches the database. Connections use HTTPS with TLS in transit. Passwords are bcrypt-hashed and API keys are stored as SHA-256 digests.
Independently verified
Security you can verify yourself
Don't take our word for it — click any badge to run the test yourself.
Security headers
securityheaders.com
CSP · HSTS · X-Frame-Options · Referrer-Policy + 3 more
Verify yourselfAuth Hardened
Independent dual-AI review + human verification
Authentication, tenant isolation, rate limits, and tool safety reviewed before submission.
Read report