PROJECT COMMAND
TWIX Lock protection overview.
The dashboard is intentionally operational: source state, model coverage, simulations, private findings, authorization and continuous watch are visible without exposing vulnerability payloads.
Executable twin lifecycle
Current target state flows through the same protected pipeline on every approved update.
CONTINUOUS
01Authorizerepo + TWIX tx
02Snapshotcommit + tree hash
03Buildisolated worker
04Reconstructactors + state
05Modelinvariants
06Challengeadversarial search
07Watchbehavioral drift
Latest protected activity
Operational metadata only. Vulnerability evidence remains sealed.
Authorization commitment
The target, source state and security authority are cryptographically bound.
REPOSITORY
—COMMIT
—SECURITY AUTHORITY
—AUTHORIZATION
—TWIX TRANSACTION
—LOCK MODEL
What the software actually is.
Lock reconstructs the implementation into an executable system map: components, authorities, asset flows, dependencies and invariants that can be tested independently of prose documentation.
Behavior graph
Demo topology reconstructed from an example Cosmos/EVM financial protocol.
User Intentwallet / tx boundary
Asset Ledgerbalances + accounting
Price Dependencyfreshness assumption
Risk Engineinvariants + limits
Modeled invariants
Examples are deliberately abstract in the public demo and contain no actionable vulnerability detail.
| ID | Invariant | Domain | Coverage | State |
|---|---|---|---|---|
INV-041 | Asset accounting remains conserved across authorized transitions. | assets | 97% | MONITORED |
INV-047 | Privileged execution requires an authority present in the current model. | authority | 94% | DEGRADED |
INV-052 | External price state satisfies configured freshness assumptions. | dependency | 88% | WATCH |
INV-063 | Failure ordering does not create a second authoritative state. | state | 91% | VERIFIED |
LOCK SIM
Challenge the model in isolation.
Simulations run only against the authorized sandbox twin. This demo shows safe scenario categories and state-search telemetry without disclosing exploit steps.
Demo estimate only. Production pricing is policy/configuration, not hard-coded here.
IDLE0%
PRIVATE FINDINGS
Verified problems. Zero public disclosure.
Finding metadata is visible to authorized project users; reproduction evidence and technical payloads remain encrypted and access-controlled.
Finding register
Open a row to view the safe private-report summary and cryptographic commitment.
| Finding | Severity | Component | Status | Commit | Action |
|---|
LOCK WATCH
Monitor behavior, not just commits.
When source, dependencies, deployments or governance change, Lock asks what changed in the executable model and which prior assumptions need to be re-tested.
Live protected event stream
Demo feed cycles through repository, model and monitoring events.
LOCK SHIELD
Pre-authorized defensive policy.
Lock never lets an LLM arbitrarily pause a financial system. A defensive action requires a deterministic verified condition and a project-authored policy that already permits the response.
GREEN
Normal operation
Continuous model and runtime observation. No intervention. Collect evidence and establish expected behavior.
YELLOW / ORANGE
Verified degradation
Alert authorized maintainers, increase simulation depth and optionally reduce configured exposure limits if the project has pre-authorized that action.
RED
Reproducible critical violation
Only a deterministic condition plus explicit project policy can invoke an emergency adapter such as pausing a risky route while preserving repayment/recovery operations.
Policy boundary
The sequence is intentionally separated.
01AI hypothesisnon-authoritative discovery
02Model conditionmachine-readable invariant
03Reproducedeterministic sandbox evidence
04Policy matchpre-authorized response scope
05Human gatewhen policy requires approval
06Adapternarrow defensive action
07Attestprivate evidence commitment
LOCK RAIL
Encrypted custody for security intelligence.
Finding packages remain private while cryptographic commitments prove integrity, chronology and remediation linkage. Access can be granted or re-keyed without public disclosure.
ENCRYPTED FINDING PACKAGE
R-000184 · Access-controlled
The underlying evidence includes the model failure, repeatable sandbox evidence, state delta, affected target state and remediation guidance. This public demo intentionally never renders the actionable payload.
REPORT COMMITMENT
7e84a2929c7c2e0b8e67b0c4e92f…CIPHERTEXT OBJECT
reverse://private/7f8419bd…ACCESS EPOCH
12REMEDIATION LINK
pendingAuthorized recipients
Demo re-keying model inspired by Securail's recipient-bound custody philosophy.
SA
Security Authoritytwix / EVM signer
OWNERLE
Lead Engineerreverse-primary-dev-2026
ACCESSLOCK BOUNTY
Private research, independently verified.
Researchers submit sealed evidence against authorized targets. Lock reproduces the claim in the project's sandbox before any project payout or severity decision proceeds.
Private submission queue
No report body or proof-of-concept is exposed in this register.
| Submission | Claim | Researcher | Stage | TWIX stake |
|---|---|---|---|---|
RB-2081 | authority invariant | researcher_0x91… | REPRODUCING | 2,500 |
RB-2079 | dependency assumption | researcher_twix1… | TRIAGE | 0 |
RB-2064 | state ordering | researcher_0x34… | PAID | 1,000 |
TWIX COMPUTE
Meter the work that creates security value.
Rather than inventing artificial token utility, Lock consumes TWIX for real platform resources: compute, simulation, monitoring, report custody, bounty settlement and verification.
Usage ledger
Illustrative demo units only; final economics can be configured at launch.
| Workload | Meter | Quantity | TWIX | Status |
|---|---|---|---|---|
| Lock Inspect | analysis worker-minutes | 428 | 2,140 | SETTLED |
| Lock Sim | sandbox state executions | 84,219 | 7,950 | SETTLED |
| Lock Watch | continuous coverage | 30 days | 2,250 | SETTLED |
| Lock Rail | encrypted report custody | 3 active | 500 | SETTLED |