About Mechanical Advantage

Mechanical Advantage is building the standard toolkit for AI agents. We give agents safe, structured, token-efficient access to the systems people use every day — web, email, calendar, contacts, and messaging — so builders can focus on what their agents think, not how they connect.

Why We Exist

AI agents are getting good at thinking. They’re still figuring out doing.

Every agent builder ends up writing the same brittle glue code to bridge the gap between what the model wants to do and the systems it needs to touch. Every useful thing an agent does comes down to a small set of verbs — browse, search, send, manage, remember — and every framework implements these primitives from scratch, with different interfaces, with security as an afterthought.

We provide these primitives as a single, clean CLI that any agent can call. Our tools are non-destructive by design — an agent with full access cannot cause irreversible data loss. Every outbound action is queued for human approval, secured by biometric passkeys that agents cannot bypass. The agent gets maximum capability with minimum blast radius.

Our Principles

• Safe by construction. Agent mistakes are reversible, not catastrophic. No deletes — only archives. No overwrites — only versions.
• Token-efficient by design. Every command returns clean, minimal output. The less context spent on tool output, the more the agent can spend on thinking.
• Standards-based. We implement open protocols — IMAP, SMTP, CalDAV, CardDAV — authenticated via OAuth 2.0. The provider is always the source of truth.
• Human-in-the-loop. Every outbound action requires biometric approval before execution. The review queue isn’t just a safety gate — it’s a teaching loop.

Our Team