Frequently asked questions
Straight answers about the CLI, models, pricing, and where your code goes. If you don't see yours, get in touch.
What is Memcode?+
Memcode is an agentic coding CLI with persistent memory of your codebase. It keeps a living picture of your repo, including what was decided and why, and runs a coding agent in your terminal on top of that memory. Every session starts already oriented instead of re-reading the repo. The Memcode web app adds chat with every frontier model, custom scheduled agents, and DataHub, a knowledge base your team can chat against.
How is it different from other coding agents?+
Most coding agents rebuild context from scratch every session. They grep, read files, and burn tokens re-learning what they knew yesterday, and they still miss the reasoning behind the code. Memcode remembers your repo across sessions: the structure, the decisions, and the history that explains them. The longer it works in your codebase, the sharper it gets.
Which models does it support?+
OpenAI GPT-5.6 Sol, Terra, and Luna. Anthropic Claude Fable 5.1, Opus 5, Sonnet 5, and Haiku 4.5. Google Gemini 3.1 Pro, 3.6 Flash, and 3.5 Flash Lite. xAI Grok 4.6. Kimi K3 and GLM 5.2 through Fireworks. Pick one with /model and the whole session runs on it; your choice is remembered per project.
How does pricing work?+
The Free plan runs everything — the CLI and the web app — on your own provider API keys, with no card and no credits. Paid plans add hosted models with prepaid credits, so no keys are needed: Personal is $25 per month, and Team is $100 per month with 5 seats included plus $20 per month for each additional seat. Paid plans include a one-time $5 starter credit to get you going. Model usage is billed at exact provider cost, with no markup.
Can I bring my own API keys?+
Yes. Add your own provider keys and Memcode uses them first, and turns served on your keys are never debited from your credit balance. Keys are encrypted at rest in a dedicated secrets vault, are never written to the application database or logs, and can only be read by the service that routes your requests. If a key fails, the turn fails cleanly rather than silently falling back to metered credits. BYOK works on every plan — the Free plan runs on it entirely.
Where does my code go?+
Your code and its memory live on your machine. When the agent needs a model, the relevant context for that request is sent through the Memcode gateway to the model provider. The gateway is stateless and holds prompts only in transit. The web app stores only what you put in it: chats, agent runs, and documents you upload to DataHub.
Do you train AI models on my code or data?+
No. We do not use your code, conversations, files, or any other data to train AI models, ours or anyone else’s. The model providers we route to (OpenAI, Anthropic, Google, xAI, Fireworks) process your requests under enterprise API terms that prohibit training on your data.
Do you sell my data to third parties?+
No. We do not sell or rent your data to anyone. Your data is shared only with the AI model provider serving your request, and only for the purpose of serving it. No advertisers, no data brokers, no third-party analytics resale.
Is Memcode SOC 2 compliant?+
Memcode runs entirely on SOC 2 Type II certified infrastructure providers (Vercel, Supabase, Google Cloud), and we maintain a security program aligned with NIST SP 800-171, with our System Security Plan, POA&M, and incident response plan published at memcode.ai/security. A SOC 2 Type II audit of Memcode itself is on our roadmap and tracked openly in the POA&M.
What platforms does the CLI run on?+
macOS, Linux, and Windows. It ships as a single static binary with a fast terminal UI. No Electron, no IDE plugin, no runtime dependencies.
Have a different question? Contact us →