Agentic proof boundary

Govern whether autonomous work is allowed to count.

AutoKirk does not replace your agents, automations, or systems. It governs whether their claims are allowed to become business truth.

Claim sourceAuthority boundaryProof evaluationReceipt rationale

What AutoKirk can now govern

Human, API, automation, and agent claims.

AutoKirk now supports governed provenance for:

  • human work claims
  • API-triggered actions
  • automation workflows
  • AI agents
  • multi-agent systems
  • external system assertions

Authority-aware completion

Approve. Deny. Conditional.

AutoKirk evaluates whether claimed work should count before closure.

  • authority boundaries
  • machine-readable rationale
  • cited controls
  • required follow-up
  • receipt-backed closure
  • append-only proof trails

Governed flow

How the proof boundary works.

01

Register claim source

Identify who or what is making the claim.

02

Define authority boundary

Define what the source may claim or close.

03

Attach obligation context

Bind the claim to a governed obligation.

04

Evaluate proof

Approve, deny, or conditionally gate completion.

05

Emit receipt

Preserve rationale, controls, and proof history.

Live governed APIs

Operational proof-boundary surfaces are now live.

api.register_claim_source(...)
api.upsert_authority_boundary(...)
api.attach_obligation_claim_context(...)
api.evaluate_proof_boundary(...)

These APIs operate beside the existing governed obligation kernel and canonical receipt path.