Governed AI. Provable decisions.
Open source.
Every agent gets a cryptographic identity. Every action leaves a tamper-proof trail. Every policy enforced at runtime, not in code reviews.
Forked & extended by engineers at
AI agents ship without governance infrastructure.
Frameworks help you build agents. None of them ship identity, policy enforcement, or provable audit trails. That gap grows with every agent you deploy.
No identity
Shared API keys. No way to verify which agent made which call.
No decision trail
No reasoning trace. No chain of custody. When something fails, you reconstruct from logs.
No proof
Logs can be edited. There is no cryptographic record of what happened or who authorized it.
Four pillars of production trust
Cryptographic Identity
Every agent gets a unique cryptographic identity. Not API keys, not tokens. Immutable, auditable, unforgeable.
Cryptographic agent identity with DID verification and credential binding
Active
control-plane.agentfield.dev
Mar 23, 2026
Apr 23, 2026
Runtime Governance
Policies enforced at the infrastructure layer, not just in code reviews. What agents can access, who they can call, what data they can touch.
Runtime policy enforcement with real-time decision logging
Decision Traceability
Trace every decision chain, every delegation, every reasoning step. Not just metrics. Full provenance of what the agent considered, decided, and why.
Full decision tracing with reasoning chain capture and provenance
Provable Audit Trail
Tamper-proof chain of every action. Who did what, when, under whose authority. Cryptographic proof, not just logs.
Merkle-tree verified audit trails with tamper-proof logging
Wherever agents need to be accountable
If your industry requires proof of what AI did and why, AgentField was built for you.
Financial Services
Tamper-proof audit trails for every trading decision. Prove compliance to regulators with cryptographic evidence.
Healthcare
Cryptographic provenance for clinical AI. Know which agent accessed what patient data, when, and under whose authority.
Supply Chain
Governed agent orchestration across vendors. Every handoff traced and verifiable across organizational boundaries.
Security & Defense
Zero-trust agent identity with runtime policy enforcement. Full provenance for threat detection and response pipelines.
Insurance
Provable claims processing. Every AI decision carries a cryptographic audit chain from intake to resolution.
Legal & Compliance
Full decision traceability for document review agents. Prove who reviewed what, when, and under whose authority.
+ Government, Energy, Manufacturing, Telecommunications, and more
Enterprise Capabilities,
Built In
Canary deployments, fleet management, durable execution, and compliance export. Production infrastructure for agent operations at scale.
Canary Deployments
Roll out agent versions 5% to 50% to 100%. A/B test agents. Blue-green deploys. Zero downtime.
Fleet Management
Remote agent management via Connector API. Air-gapped deployment. Bearer token authentication.
Compliance Export
SOC2, HIPAA, regulatory. Export audit chains for any compliance framework with a single API call.
Human-in-the-Loop
One call suspends execution for human approval. Crash-safe, durable, with webhook notifications.
Durable Async Execution
No timeout limits. Agents run for hours or days. HMAC-signed webhooks on completion. Auto-retry with backoff.
Self-Hosted / On-Prem
Apache 2.0. Run on your infrastructure with full control.
Open source. Audit the code yourself.
Every line under Apache 2.0. Read the cryptography. Verify the governance logic. Self-host on your infrastructure.
1.1k
GitHub Stars
v0.1.62
Latest Release
21
Commits / Week
165
Forks
Governed AI. Provable decisions.
Start today.
See how AgentField gives your agents identity, governance, and provable audit trails from day one.