Everything stays
on your machine
Magnum is designed for developers who expect their work, code, and API keys to remain private. The app runs locally. Keys stay in the system keychain. No project data is shared with our servers.
Commitments
No project data collection
No code, prompts, or file contents collected. Telemetry is limited to anonymous usage data and can be turned off in Settings. What you build stays between you and your machines.
Keys stay in your keychain
Provider keys live in the system keychain (Keychain on macOS, Secret Service on Linux). Magnum reads one when a session starts and hands it to that session's agent, on whichever of your machines it runs. Never sent to us, synced, or cached elsewhere.
Code is yours
Project files, repositories and your agents' transcripts stay on your machines. Not mirrored, uploaded, or inspected.
Local-first architecture
A native Rust app, not Electron; on macOS a web view shows only the preview of your own app when you run it. Its node on each machine listens on a local socket only your user can open, never on a network port, and your other machines are reached over SSH. No cloud backend relaying or storing prompts.
Implementation
Local processing only
Magnum runs each agent's own program (Claude Code, Codex or Pi) on your computer or on a machine you added, and talks to it directly. The agent sends your prompts and the code it reads to its model provider, as it does in a terminal. Never proxied through our services.
No surprise storage
No hidden caches or background sync jobs. Your agents keep running after Magnum quits, in Magnum Helper (Magnum's node on Linux) and on your other machines, and Magnum picks them up when it opens again; the Helper stops by itself once nothing runs in it and Magnum is closed. Magnum keeps its settings and Magnum Router's decision log; sessions are each agent's own history.
Transparent permissions
Permission rules and approval prompts run entirely on your computer. Full control over what tools can touch which files.
Questions
Do you upload any part of my codebase or my agents' history?
No. Magnum neither uploads nor mirrors repositories or prompts: they go only to your machines and to the model provider each agent uses. Magnum itself checks GitHub for its own updates, asks pi.dev for Pi's newest version, and sends one usage event as it opens while telemetry is on.
Are API keys stored on Magnum servers?
Never. Keys you add in Magnum, Codex's OpenAI key among them, are kept in the system keychain (Keychain on macOS, Secret Service on Linux), and each agent keeps its own sign-in. Magnum hands a key only to the agent a session runs, on whichever of your machines it runs.
Can I opt out of telemetry?
Yes. Telemetry is completely optional and can be disabled in Settings. We collect only anonymous usage data (device ID, platform, hardware specs, app version) to improve the product. No personally identifiable information. No file contents. Toggle it off anytime instantly.
How to reach you?
Email magnum@prayukta.com for privacy feedback or help auditing setup.