Personal AI agents you actually own.
An always-on agent that lives on your machine, thinks with the model you choose, and works while you're away. It's built so it cannot see your keys, read your files, or reach the internet unless you granted exactly that.
How it works
Isolation is built in, not bolted on.
Most agent frameworks start with a powerful agent and add guardrails around it. Hatchling starts with the boundary and gives the agent only what you grant.
Every agent runs in a cell
Its own user, mount, PID, network and IPC namespaces, a seccomp-BPF syscall allowlist, and every capability dropped. There is no unsandboxed mode: no cell, no agent.
Your keys never enter the cell
Credentials are sealed at rest (XChaCha20-Poly1305). At the boundary, the egress proxy swaps a broker-minted surrogate for the real key, so the agent can use a service without ever holding its key.
One way out
A cell's network namespace is empty. It reaches the outside only through the broker's proxy, per grant, against an exact-match allowlist with pinned addresses and TLS hostname checks.
Held to a high bar
Audited before launch, and the gaps are public.
Before going public, Hatchling's trust boundary was reviewed: who may connect, egress, the vault, and cell isolation. No way out of a cell was found. Every gap that was found is published with a work order that's open to contributors.
unsafe; every other crate denies itLicensing
Free for everyone. Commercial licenses for teams.
Open source: AGPL-3.0-or-later
Use it, study it, change it, run it. If you offer a modified Hatchling to others over a network, you share your changes under the same license.
Commercial license
For companies that need terms the AGPL doesn't give them:
- private modifications and closed integrations
- custom builds for your environment
- seat or site licensing, and support