Flagship artifact
Forge rescue suite
The rescue suite is the clearest current example of evidence-governed computation: replayable trace, named obligation, concrete witness, restricted semantic theorem, approval gate, and public claim boundary.
Evidence Browser / v0
Inspect computational claims and their boundaries.
Every claim Monogate ships as an evidence bundle — traces, replay logs, proof obligations, semantic tiers, and explicit non-claims. This page shows what has been approved for a surface and what remains candidate evidence.
validationreplaysemantic strengthclaim boundaryoptimizer evidence
Forge Rescue Suite
approved_for_surface1 restricted rewrite, 3 concrete invariantsBoundary-event rescue packets with replay validation, MachLib witness routing, and a restricted log-domain semantic theorem.
simulated hardware evidence
Electronics Trainer Packet
approved_for_surfaceevidence grammar packetSimulated Trainer Board packet using the shared evidence grammar for trace, replay, validation, and claim flags.
runtime evidence
Monogate OS Replay Chain
candidate_onlycandidate replay chainReplay-native runtime evidence with identity CI, positive/negative fixture gates, and Observatory status; displayed as candidate evidence only.
capability evidence
CapCard Internal Lane
candidate_onlyinternal candidate capabilityInternal CapCard marketplace/import evidence that exercises review grammar without public-ready promotion.
agent evidence
Agent Output Evidence Demo
candidate_onlyclaim governed agent outputA bounded agent answer wrapped in validation, replay, claim flags, non-claims, and reviewer decision metadata.
runtime evidence
Monogate OS EML Bridge
candidate_onlyeml shaped fixture adapter qemu replay sensitivityCandidate EML/Forge-shaped OS bridge fixture set: log-domain lift and guard-clamp kernels run through mirrored no_std adapters with QEMU replay and sensitivity evidence.
Evidence index
| artifact | decision | validation | replay | semantics | claim boundary |
|---|---|---|---|---|---|
| Forge Rescue Suite | approved_for_surface | pass | pass | 1 restricted rewrite, 3 concrete invariants | Approved for existing public/dev surfaces; no full optimizer semantic rewrite or hardware observation is claimed. |
| Electronics Trainer Packet | approved_for_surface | pass | pass | evidence grammar packet | Approved as simulated evidence only; hardware truth requires a real capture packet from the laptop-agent lane. |
| Monogate OS Replay Chain | candidate_only | pass | pass | candidate replay chain | Candidate only; not a bootable OS claim, not certified safety, and not production runtime assurance. |
| CapCard Internal Lane | candidate_only | pass | pass | internal candidate capability | Candidate only; no public marketplace promotion and no external capability certification. |
| Agent Output Evidence Demo | candidate_only | pass | pass | claim governed agent output | Candidate only; demonstrates claim-governed agent output, not general agent truthfulness or autonomous deployment readiness. |
| Monogate OS EML Bridge | candidate_only | pass | pass | eml shaped fixture adapter qemu replay sensitivity | Candidate only; these are fixture adapters, not Forge compiler outputs, not a Forge OS target, and not a bootable OS claim. |
Evidence paths
forge/reports/proof_carrying_rescue_suite_v0_2026_05_26.jsonforge/reports/proof_carrying_rescue_replay_v0_2026_05_26.jsonforge/reports/rescue_obligation_registry_v0_2026_05_26.jsonforge/reports/rescue_artifact_approval_v0_2026_05_26.jsonNew frontier
Claim-governed agents
Agent outputs are now represented as candidate evidence packets with task, output, replay log, validator checks, claim flags, and explicit non-claims.
Open agent packet