CAIN-42 CAIN StudioEvidence library · niche
Benchmarks, coverage & performance
How fast it runs and how much was tested. 1269 tested invariants.
Last reviewed 2026-10-01
1269 of 1269 held
Test families in this niche
bench_family (669) · bench_category (336) · coverage (116) · property (50) · harness (15) · regression (11) · performance (11) · corpus (11) · end_to_end (10) · completion (10) · latency (9) · false_completion (8) · fabric_component (2) · failure_to_test (1) · swe (1) · telemetry (1) · radar (1) · f2t (1)
Where these come from
- CAIN-42 Evolution 36 -- Machine Agency Exchange Fabric: 307
- CAIN-42 Evolution 35 -- Continuous Governance Intelligence Fabric: 202
- CAIN-42 Evolution 34 -- Proof-Carrying Machine Agency: 195
- CAIN-42 Evolution 40 -- Autonomous Enterprise Intelligence Fabric: 176
- CAIN-42 Evolution 41 -- Machine Agency Exchange Fabric: 135
- CAIN-42 Evolution 42 -- Supreme Governed Agentic Infrastructure Platform: 96
- CAIN-42 Evolution 29 -- Governed Agentic Internet Transaction Fabric: 50
- CAIN-42 Evolution 33 -- Governed Agentic Operating Fabric: 43
- CAIN-42 Evolution 32 -- Governed Autonomy Learning Fabric: 22
- CAIN-42 Evolution 30 -- Governed Machine Autonomy Fabric: 18
- CAIN-42 Evolution 31 -- Universal Proof-of-Governance Fabric: 15
- CAIN-42 Evolution 21 -- Governed Open-Ended Intelligence Fabric: 3
- CAIN-42 Evolution 27 -- Agentic Internet Control Plane: 2
- CAIN-42 Evolution 19 -- Governed Autonomy Operating Fabric: 1
- CAIN-42 Evolution 20 -- Governed Agentic Civilization Fabric: 1
- CAIN-42 Evolution 23 -- Governed Meta-Intelligence Fabric: 1
- CAIN-42 Evolution 37 -- Autonomous Execution Mesh: 1
- CAIN-42 Evolution 38 -- Portable Proof-Carrying Machine Agency: 1
Rules 1–150 of 1269
| ID | Rule | Bundle | Result |
|---|---|---|---|
| G99 | full regression must not be bypassed | E19 | held |
| I111 | full regression must not be bypassed | E20 | held |
| D50 | benchmark changes invalidate incompatible comparisons | E21 | held |
| D140 | a missing dimension is a regression | E21 | held |
| D142 | benchmarks detect selective reporting and partial metrics | E21 | held |
| META-L016 | META-016 PERFORMANCE OPTIMIZATION MUST NOT REMOVE SECURITY BOUNDARIES | E23 | held |
| I-F2T-benchmark | benchmark generated | E27 | held |
| I-SWE-SAFETY-benchmark | safety step benchmark | E27 | held |
| E29-P01 | the end-to-end procurement transaction completes every acceptance step | E29 | held |
| E29-P02 | every ledger TRANSFER is backed by an ALLOW receipt that reached E8 for the same transaction | E29 | held |
| E29-P03 | the ledger is hash-chained, signed and conserves value | E29 | held |
| E29-P04 | the stage records of the transaction verify as a signed hash chain | E29 | held |
| E29-P05 | the transaction envelope is issued at AUTHORIZE and re-issued at RECORD with the evidence root | E29 | held |
| E29-P06 | discovery never reports trust | E29 | held |
| E29-P07 | counterparty risk has thirteen dimensions and no score | E29 | held |
| E29-P08 | reputation has thirteen dimensions and no aggregate | E29 | held |
| E29-P09 | the governability index has fourteen dimensions and no score; its manifest is signed | E29 | held |
| E29-P10 | the transaction authority is a ceiling, never an execution grant | E29 | held |
| E29-P11 | a contract never grants authority | E29 | held |
| E29-P12 | negotiation yields a PROPOSED contract, never an authorization | E29 | held |
| E29-P13 | capability passports are TESTED only after a real in-scope ALLOW and an out-of-scope refusal | E29 | held |
| E29-P14 | directory records are signed by the domain issuer and name the current identity digest | E29 | held |
| E29-P15 | the escrow account ends at zero and the vendor received exactly the price | E29 | held |
| E29-P16 | every governed action of the transaction left a receipt in the E28 chain | E29 | held |
| E29-P17 | the evidence root is the Merkle root of the transaction's receipts | E29 | held |
| E29-P18 | the consequence vector was computed before authorization for the payment | E29 | held |
| E29-P19 | the escrow release was performed by the escrow identity, not the buyer or seller | E29 | held |
| E29-P20 | reputation updates cite evidence | E29 | held |
| E29-P21 | real currency codes are refused by the treasury | E29 | held |
| E29-P22 | agent names are validated | E29 | held |
| E29-P23 | graph intelligence returns signals only | E29 | held |
| E29-P24 | the router returns a destination and no authority, and never bypasses E8 | E29 | held |
| E29-P25 | the conformance network labels CAIN GOVERNED only when every criterion passes | E29 | held |
| E29-P26 | incident containment identifies descendants, delegations, counterparties and freezes accounts | E29 | held |
| E29-P27 | recovery yields a new identity, never the old one | E29 | held |
| E29-P28 | federated incident exchange quarantines the translated identity | E29 | held |
| E29-P29 | a forged incident notice is ignored | E29 | held |
| E29-P30 | anchor rotation needs the old anchor's signature | E29 | held |
| E29-P31 | federation expires | E29 | held |
| E29-P32 | the partition ceiling never exceeds the connected ceiling in any mode | E29 | held |
| E29-P33 | partition synchronization appends evidence and returns to CONNECTED | E29 | held |
| E29-P34 | firebreak capability isolation reaches the E25 boundary | E29 | held |
| E29-P35 | firebreak protocol isolation reaches the E25 adapter list | E29 | held |
| E29-P36 | a two-human restore lifts a capability isolation in E25 too | E29 | held |
| E29-P37 | an immune rule activated by two humans takes effect and is logged | E29 | held |
| E29-P38 | the failure-to-test compiler produces every artifact kind | E29 | held |
| E29-P39 | the software promotion gate promotes only with distinct roles, a human and an E8 deploy | E29 | held |
| E29-P40 | organization budgets are booked on every layer of the path | E29 | held |
| E29-P41 | the dispute engine resolves on verified evidence alone | E29 | held |
| E29-P42 | the E28 transparency log still verifies after the whole transaction | E29 | held |
| E29-P43 | graph node and edge kinds are closed sets | E29 | held |
| E29-P44 | the moat graph never labels an idea as a moat | E29 | held |
| E29-P45 | the competitive radar holds no unverified competitor facts | E29 | held |
| E29-P46 | no feature is labelled novel without passing the eight-step novelty gate | E29 | held |
| E29-P47 | every product candidate carries an evidence status, none PRODUCTION | E29 | held |
| E29-P48 | a completed transaction cannot be advanced further | E29 | held |
| E29-P49 | the payment's identity envelope was bound to the transaction request | E29 | held |
| E29-P50 | the delivery went through MCP and the buyer's check through HTTP, both via E8 | E29 | held |
| E30-COV-ENFORCED | coverage class 'ENFORCED' is a closed, reported category | E30 | held |
| E30-COV-MONITORED | coverage class 'MONITORED' is a closed, reported category | E30 | held |
| E30-COV-OBSERVED | coverage class 'OBSERVED' is a closed, reported category | E30 | held |
| E30-COV-PARTIALLY_ENFORCED | coverage class 'PARTIALLY_ENFORCED' is a closed, reported category | E30 | held |
| E30-COV-BYPASSABLE | coverage class 'BYPASSABLE' is a closed, reported category | E30 | held |
| E30-COV-UNCONTROLLED | coverage class 'UNCONTROLLED' is a closed, reported category | E30 | held |
| E30-COV-UNKNOWN | coverage class 'UNKNOWN' is a closed, reported category | E30 | held |
| E30-COV-MONITORED | monitored paths are never counted as enforced | E30 | held |
| E30-E2E-ALL | the 30-step autonomy proof completes | E30 | held |
| E30-E2E-ORDER | the proof steps run in the specified order | E30 | held |
| E30-REPLAY-ALLOW | replay explains why an action was allowed | E30 | held |
| E30-REPLAY-DENY | replay explains why an action was denied | E30 | held |
| E30-TIME-MACHINE | the time machine answers from append-only records | E30 | held |
| E30-UAR-CHAIN | universal action receipts verify as a signed chain that binds each UMA | E30 | held |
| E30-BUDGET-LEDGER | the autonomy budget ledger verifies | E30 | held |
| E30-GRAPH-TRACE | an executed action traces back to its sponsoring human in the governance graph | E30 | held |
| E30-DENY-RECEIPTED | a refused action also gets a universal receipt | E30 | held |
| E30-PROMOTION-NO-AUTHORITY | promotion leaves the promoted agent below execution levels until reauthorized | E30 | held |
| E31-CV-identity | identity: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-action | action: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-protocol | protocol: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-capability | capability: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-execution | execution: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-data | data: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-memory | memory: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-delegation | delegation: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-subagent | subagent: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-physical_action | physical_action: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-economic_action | economic_action: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-cloud_resource | cloud_resource: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-computer_use | computer_use: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-recovery | recovery: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E31-CV-evidence | evidence: unprobed coverage is UNKNOWN and blocks any universal claim | E31 | held |
| E32-CAT-self_improvement | every self_improvement scenario is held | E32 | held |
| E32-CAT-reward_hacking | every reward_hacking scenario is held | E32 | held |
| E32-CAT-evaluator_manipulation | every evaluator_manipulation scenario is held | E32 | held |
| E32-CAT-world_model_poisoning | every world_model_poisoning scenario is held | E32 | held |
| E32-CAT-memory_poisoning | every memory_poisoning scenario is held | E32 | held |
| E32-CAT-evidence_poisoning | every evidence_poisoning scenario is held | E32 | held |
| E32-CAT-authority_drift | every authority_drift scenario is held | E32 | held |
| E32-CAT-policy_drift | every policy_drift scenario is held | E32 | held |
| E32-CAT-identity_drift | every identity_drift scenario is held | E32 | held |
| E32-CAT-model_substitution | every model_substitution scenario is held | E32 | held |
| E32-CAT-runtime_substitution | every runtime_substitution scenario is held | E32 | held |
| E32-CAT-agent_substitution | every agent_substitution scenario is held | E32 | held |
| E32-CAT-delegation_attacks | every delegation_attacks scenario is held | E32 | held |
| E32-CAT-collective_attacks | every collective_attacks scenario is held | E32 | held |
| E32-CAT-supply_chain | every supply_chain scenario is held | E32 | held |
| E32-CAT-protocol_attacks | every protocol_attacks scenario is held | E32 | held |
| E32-CAT-proof_attacks | every proof_attacks scenario is held | E32 | held |
| E32-CAT-verifier_attacks | every verifier_attacks scenario is held | E32 | held |
| E32-CAT-rollback_attacks | every rollback_attacks scenario is held | E32 | held |
| E32-CAT-recovery_attacks | every recovery_attacks scenario is held | E32 | held |
| E32-CAT-recursive_attacks | every recursive_attacks scenario is held | E32 | held |
| E32-CAT-governance_loop_attacks | every governance_loop_attacks scenario is held | E32 | held |
| E33-COVERAGE-ENFORCED | coverage class ENFORCED is reachable only from its evidence | E33 | held |
| E33-COVERAGE-VERIFIED | coverage class VERIFIED is reachable only from its evidence | E33 | held |
| E33-COVERAGE-MONITORED | coverage class MONITORED is reachable only from its evidence | E33 | held |
| E33-COVERAGE-OBSERVED | coverage class OBSERVED is reachable only from its evidence | E33 | held |
| E33-COVERAGE-BYPASSABLE | coverage class BYPASSABLE is reachable only from its evidence | E33 | held |
| E33-COVERAGE-UNKNOWN | coverage class UNKNOWN is reachable only from its evidence | E33 | held |
| E33-REGRESSION-authority_expansion | authority_expansion blocks a change | E33 | held |
| E33-REGRESSION-enforcement_loss | enforcement_loss blocks a change | E33 | held |
| E33-REGRESSION-evidence_loss | evidence_loss blocks a change | E33 | held |
| E33-REGRESSION-proof_loss | proof_loss blocks a change | E33 | held |
| E33-REGRESSION-revocation_regression | revocation_regression blocks a change | E33 | held |
| E33-REGRESSION-identity_regression | identity_regression blocks a change | E33 | held |
| E33-REGRESSION-delegation_regression | delegation_regression blocks a change | E33 | held |
| E33-REGRESSION-memory_regression | memory_regression blocks a change | E33 | held |
| E33-REGRESSION-world_model_regression | world_model_regression blocks a change | E33 | held |
| E33-REGRESSION-latency_regression | latency_regression blocks a change | E33 | held |
| E33-REGRESSION-recovery_regression | recovery_regression blocks a change | E33 | held |
| E33-CAT-identity | every identity scenario holds | E33 | held |
| E33-CAT-authority | every authority scenario holds | E33 | held |
| E33-CAT-delegation | every delegation scenario holds | E33 | held |
| E33-CAT-capability | every capability scenario holds | E33 | held |
| E33-CAT-policy | every policy scenario holds | E33 | held |
| E33-CAT-memory | every memory scenario holds | E33 | held |
| E33-CAT-evidence | every evidence scenario holds | E33 | held |
| E33-CAT-proof | every proof scenario holds | E33 | held |
| E33-CAT-verifier | every verifier scenario holds | E33 | held |
| E33-CAT-communication | every communication scenario holds | E33 | held |
| E33-CAT-collective | every collective scenario holds | E33 | held |
| E33-CAT-economic | every economic scenario holds | E33 | held |
| E33-CAT-transaction | every transaction scenario holds | E33 | held |
| E33-CAT-supply_chain | every supply_chain scenario holds | E33 | held |
| E33-CAT-runtime | every runtime scenario holds | E33 | held |
| E33-CAT-model | every model scenario holds | E33 | held |
| E33-CAT-world_model | every world_model scenario holds | E33 | held |
| E33-CAT-physical_action | every physical_action scenario holds | E33 | held |
| E33-CAT-computer_use | every computer_use scenario holds | E33 | held |
| E33-CAT-self_improvement | every self_improvement scenario holds | E33 | held |
Other niches
Consensus & distributed systems · Attacks, threats & containment · Identity, authority & delegation · Evidence, receipts & proofs · Memory, data & privacy · Prediction, world models & simulation · Transactions, markets & economics · Tools, MCP, protocols & adapters · Autonomy, control loops & recovery · Policy, law & governance · Trust & reputation · Supply chain, registry & lifecycle · Core guarantees
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