CAIN-42 CAIN StudioCAIN-42 evidence library
CAIN-42 Evolution 33 -- Governed Agentic Operating Fabric
Evidence bundle for evolution E33: 6 claims, 581 of 581 invariants held.
Last reviewed 2026-10-01
Browse the raw bundle · How to reproduce it · SHA-256 manifest
Claims (6)
| Claim | Level |
|---|---|
| C42-E33-OPERATION: every consequential operation kind runs one explicit 16-state lifecycle; executed operations pass the E32 overlay, E30, E28, E25 and E8 and carry an E31 proof; state-changing operations (messages, code, subagents, governed memory) run inside sealed executors behind E8; governance mutations are routed to their governing engine and never executed | TESTED |
| C42-E33-SIDECAR: an external agent process that imports nothing from CAIN is governed through a stdio sidecar speaking a signed, versioned ABI; forged and malformed requests are refused | TESTED |
| C42-E33-LOOP: a 27-step supreme loop: an operation allowed under baseline governance is refused after E32's governed learning promotes a stricter rule | TESTED |
| C42-E33-BENCH: 2029 of 2029 scenarios across 26 categories held; 581 of 581 invariants; mutation 12 of 12 (targeted) | TESTED |
| C42-E33-HONEST-SCOPE: runtime adapters, gateway surfaces, SDK targets, cloud services and moat foundations are each classified from what exists; the cloud is NOT DEPLOYED and no market moat is claimed | TESTED |
| C42-E33-SCALE: synthetic in-process scale runs; the largest operation rows are routing decisions or event-chain hashing only | SIMULATED |
Invariants: 581 of 581 held
Rules the code must never break, each checked across many scenarios. See all 581 on one page.
| Niche | Held |
|---|---|
| Tools, MCP, protocols & adapters | 109 of 109 |
| Transactions, markets & economics | 82 of 82 |
| Policy, law & governance | 74 of 74 |
| Evidence, receipts & proofs | 64 of 64 |
| Attacks, threats & containment | 59 of 59 |
| Identity, authority & delegation | 49 of 49 |
| Benchmarks, coverage & performance | 43 of 43 |
| Core guarantees | 29 of 29 |
| Autonomy, control loops & recovery | 18 of 18 |
| Prediction, world models & simulation | 16 of 16 |
| Supply chain, registry & lifecycle | 16 of 16 |
| Trust & reputation | 11 of 11 |
| Consensus & distributed systems | 8 of 8 |
| Memory, data & privacy | 3 of 3 |
Verify it in your browser
Your browser downloads the bundle's SHA256SUMS manifest and the file(s) behind this page, hashes them with SHA-256 locally (WebCrypto), and compares. A match shows the record you are reading is the published one; it does not by itself prove who published it (see the signed claims registry and the bundle verifier for that).
Known limitations
- An in-process TESTED library; not hosted; the operation kernel is not wired into the gateway, MCPGate or the clusters. The sidecar runs locally over stdio against a reference world.
- Adapters are E25 reference adapters; no network wire implementation of any protocol; workflow engines, enterprise automation and physical/robotic systems are UNKNOWN or refused.
- The Governance Cloud is architecture only (NOT DEPLOYED); Go, REST and gRPC SDKs are NOT IMPLEMENTED.
- Code execution is limited to a whitelisted pure-function runner; arbitrary code is not sandboxed or governed.
- Communication governance is structural (declared paths, message classes, coalitions); content steganalysis and covert-channel detection inside text are NOT IMPLEMENTED.
- World-state version is not bound (UNKNOWN) and physical environments are refused rather than governed.
- Moat foundations are technical only: no customers, integrations, standards participation or market position.
- The research radar lists 11 items from one web search on 2026-09-30 (snippets only, not read in full, reproducibility not checked).
- The mutation self-test runs each mutant against the bench subset that exercises the mutated component.
- Scale rows are synthetic and in-process; 100,000 operations are routing decisions only and the 1,000,000 and 10,000,000 rows are event-chain hashing only.
The bundle's own README
CAIN-42 Evolution 33 -- Governed Agentic Operating Fabric#
Every consequential machine operation runs one governed lifecycle over the existing stack (E32 overlay -> E31 proof -> E30 -> E28 -> E25 -> E8). An external agent can be governed through a stdio sidecar without importing CAIN. Verify with verify_e33.py.txt (see REPRODUCTION.md). Status: TESTED library, pre-production.
Try CAIN-42 on your own agents
Create a free account and every new account starts with a 7-day trial of the full platform. Or try the sandbox first, with no account at all.
Create a free account → · Try the sandbox · See the whole ecosystem