Every agent action.
Signed. Chained. Verified.
Treeship gives your AI agents a cryptographic audit trail. Ed25519 signed receipts for every action, Merkle checkpoints for every session, offline verification for every stakeholder.
Signed Receipts
Ed25519 signatures on every agent action. Output digests, file changes, and git state captured in tamper-proof artifacts.
Merkle Checkpoints
Receipts chain to previous actions automatically. Generate inclusion proofs. Verify entire sessions offline with a 167KB WASM bundle.
Human Approval Gates
Flag risky actions for human review. Single-use nonce binding prevents replay. TUI dashboard for real-time session management.
OpenTelemetry Export
Stream agent telemetry to Jaeger, Datadog, or Langfuse. Decision capture records LLM selection, token counts, and confidence.
Multi-Language SDKs
First-class support in Rust, TypeScript, and Python. MCP bridge included. npm install and you're running in under a minute.
Open Source
Apache 2.0 licensed. 120+ passing tests. 30+ CLI commands. Inspect every line. No vendor lock-in, no trust required in us.
Drop into your existing agent stack
Trust is the last unsolved layer
in the agent stack.
AI agents are ready for production. The question isn't capability anymore. It's proof. Treeship makes every action verifiable, every session auditable, and every deployment trustworthy.