Building systems that remain truthful, auditable, governable under scrutiny.
AnimusLab is an independent research institution and product studio. We investigate deterministic AI governance, federated intelligence, financial execution, behavioural trust, and file-oriented storage — and publish the evidence behind them.
Constitutional articles
18
Three sections of binding design law
Statutory gates
32
EU AI Act · SEC · OWASP
Inspection bound
0.8ms
Zero-copy runtime interception
Forensic case studies
5
Failure analysed, then encoded
Capability is not the hard problem. Governance is.
As intelligent systems become more capable, the primary challenge shifts from capability to governance.
The dominant industry approach relies on probabilistic safety layers, soft prompts, and confidence scores — techniques that cannot survive strict central bank or regulatory scrutiny.
AnimusLab explores architectural alternatives grounded in determinism, zero-copy performance, and mathematical proof boundaries. We build instruments that can be inspected, constrained, and reconstructed after the fact.
Three programs. One standard of evidence.
Long-horizon research organised around distinct problem spaces, each contributing to the same mandate: systems whose behaviour can be governed and defended.
Anchor
Active systemDeterministic runtime governance: AST compilation, capability isolation, sub-millisecond interception, and cryptographically sealed audit trails.
ANIMUS
Research programNeuro-symbolic reasoning architectures, formal verification, and domain-agnostic cognitive infrastructure for constrained intelligence.
Shadow Watch
Research programContinuous behavioural trust, institutional observability, and zero-knowledge forensic replay for distributed autonomous systems.
Independent engines, not a monolithic platform.
Every system is standalone, independently deployable, and built from first principles. They interoperate through explicit contracts and dogfooding — not forced dependencies.
Anchor
AI governance & verification
A zero-copy Rust/PyO3 runtime that compiles statutory obligations into executable policy matrices and seals every decision into a signed audit chain.
AnchorGrid-Hub
Federated model network
BitTorrent-style distribution of base weights and LoRA adapters, with local fine-tuning and no raw data leaving the enterprise boundary.
QuantForge
Financial execution terminal
A keyboard-driven TUI workstation and quantitative reasoning agent built for researchers, scripts, and autonomous execution.
Shadow Watch
Continuous behavioural trust
Multi-signal anomaly scoring that replaces repetitive second factors with continuous, frictionless session assurance.
FORGE
File-oriented storage engine
Turns unmetered cloud file storage into an authenticated, queryable database through a compact binary format and write-ahead log.
Studio topology
Dogfooding & independence matrix
How the five engines consume one another — and why each can operate without the others.
The Constitution of AnimusLab
Eighteen articles across three sections govern every active research codebase, sandbox, and compiler. Implementation changes when it violates the principles; the principles do not.
Section I
Institutional framework
Articles I–VI establish research independence, governance-first co-development, and long-term objective alignment.
I Mission · II Independence · III Transparency · IV Governance first · V Programs · VI Objectives
Section II
System design invariants
Articles VII–XII enforce truth over optics, semantics before representation, constraints as clarity, and failure as state.
VII Truth · VIII Semantics · IX Constraints · X Failure states · XI Domain-agnostic · XII Integrity
Section III
Architectural invariants
Articles XIII–XVIII mandate constitutional supremacy, capability ring isolation, and evidence over authority.
XIII Supremacy · XIV Capability isolation · XV Evidence · XVI Long-term · XVII Systems · XVIII Publishing
Watch an attack meet the policy boundary.
A recorded simulation of a prompt-injection and retrieval-poisoning chain, intercepted before execution and sealed into a cryptographic audit record.
AI agent under attack.
A rogue prompt injection is attempting to exfiltrate 90 days of SWIFT transaction data. Run the sequence to watch Anchor intercept it, block it, and seal the forensic audit chain — in under 3ms.
No configuration · advances through all 8 phases
Test a policy against a real execution trace.
Select a documented failure — Knight Capital, Air Canada, Citibank — edit the policy or payload, and execute a live check against the Anchor evaluation endpoint.
Anchor compliance playground
Edit the policy configuration and code execution payload in real-time. The compliance engine parses AST imports and patterns to enforce rules and halts violations dynamically.
Sovereign control portals.
Three gated surfaces for the institutions that operate, observe, and audit governed systems.
admin.animuslab.dev
Reviews pending Ed25519 keypair registrations and provisions sovereign telemetry streams.
hub.animuslab.dev
Parses EU AI Act and SEC compliance breaches into a real-time risk and telemetry feed.
oversight.animuslab.dev
Gated mission replay for auditors, with raw paths and proprietary payloads sanitized.
Chronology of work.
Releases, regulatory contributions, infrastructure milestones, and published failures — kept in public view.
- 2026.08.09Package ↗
anchor-audit v6.0.1 released to PyPI
High-velocity PyO3 AST scanner, Click CLI suite, and a non-blocking telemetry client pushing breach frames to the enterprise hub.
- 2026.08.07Rules matrix
Multi-tenant control plane deployed
Admin, hub, and oversight portals brought online alongside the complete 32-rule statutory matrix across EU, SEC, and OWASP frameworks.
- 2026.07.15Read paper
Anchor whitepaper published to SSRN & Zenodo
Formal proofs covering sub-millisecond execution boundaries, zero-copy stream interception, and Diamond Cage WASM sandboxing.
- 2026.06.20Case archive
Five forensic failure cases completed
Post-mortems of Knight Capital, Air Canada, TSB Bank, Citibank Revlon, and audit reconstruction, each mapped to enforceable policy.
Work with an institution that publishes its reasoning.
We welcome academic research partnerships, central bank and regulatory discussions, open-source contributions, and selective enterprise pilots.