Who is Invarians for today?
Substrate awareness for DeFi risk, treasury, and CRE workflows.
For
Programmatic cross-chain operations
- DeFi risk modelers
- DAO treasury operators routing cross-chain rebalancing
- Intent solvers and keepers gating fills on bridge state
- Chainlink CRE integrators wiring structural execution gates
Out of scope
Retail, HFT, MEV
Retail daily trading, high-frequency trading, MEV extraction, micro-transfers.
North-star scope, Phase 02-03. Institutional RWA settlement at production scale requires multi-party attestation and decentralized verification. That trust model arrives with the hybrid Chainlink Functions DON deployment (Q3-Q4 2026) and the decentralized network (2027). Until then, the Phase 01 centralized pipeline is appropriate as a complementary structural signal in risk dashboards, not as a settlement gate.
Choose your trust model
Deploy at the level of trust you require.
Available · Production
Centralised
API · SDK · 1 RPC · Hetzner
pip install "invarians[requests]>=0.5.0"
panel = InvariansClient("inv_...").get_panel_v2(include="diagnostic")
br = panel.bridge_by_id("arbitrum-ethereum/cctp")
panel = InvariansClient("inv_...").get_panel_v2(include="diagnostic")
br = panel.bridge_by_id("arbitrum-ethereum/cctp")
Certified execution context
HMAC-SHA256 signed · verifiable
API + SDK · early access (Alpha, breaking changes expected)
L1 × L2 · 7 chains · 20 variable-latency bridges (CCTP calibrated, CCIP raw observable)
Trust model : Invarians server
Signing : single node
Not a decentralized proof
Q3–Q4 2026
Hybrid
Chainlink Functions · DON · Multi-RPC
chainlink.requestExecution(
source: invarians_algo,
params: calibration_v2
)
source: invarians_algo,
params: calibration_v2
)
DON threshold signature
Multi-RPC · independent nodes
Cryptographically verifiable
CRE · Chainlink Functions
Requires Chainlink CRE
Algorithm authored by Invarians
Not yet in production
2027
Decentralised
INVAR Network · Node · Stake
// Independent nodes
// Stake to compute
// Ed25519 per-node signing
// On-chain baselines
// Stake to compute
// Ed25519 per-node signing
// On-chain baselines
Independent node network
Stake to compute · node rewards
On-chain baselines · open methodology
Ed25519 per-node signing
Not yet available
Nominal Windows
What is nominal is not fixed. It evolves with the load.
As AI agents act on-chain at scale, their collective behavior modifies blockchain demand patterns, block saturation, transaction profiles, bridge posting cadence. The infrastructure baseline shifts. What was nominal yesterday may not be nominal today.
Invarians measures these shifts continuously and communicates the current nominal window, not as a static threshold, but as a live reading calibrated against each chain's own evolving distribution. The agent knows when the system is most nominal. The decision is its own.
The Logical Pipeline
1
Finalized on-chain data
Public blockchain data only. No mempool. No price feeds. No prediction.
→
2
Two measurement layers
Structural layer and demand layer, independently measured on each chain and each rollup.
→
3
State classification
Signed regime grid: 4 base codes (S1D1, S1D2, S2D1, S2D2) extended with direction suffixes (+, -, ±) on chains with calibrated lower bounds since 2026-04-29.
→
4
On-Chain Execution Context
Cross-layer infrastructure context, certified, signed, timestamped.
→
5
Agent Policy
The agent applies its own policy. The Pattern Reference maps each L1 · L2 · bridge combination to documented historical events.
Coverage
Three infrastructure layers. Independently measured.
L1, Ethereum · Polygon · Solana · Avalanche
Structural regime (τ) and demand pressure (π). 4 base states per chain (S1D1, S1D2, S2D1, S2D2) extended to 12 signed states on chains with calibrated lower bounds (since 2026-04-29).
L2, Arbitrum · Base · Optimism
Rollup execution profile, throughput, transaction composition, blob saturation. Distinct framework from L1.
Bridge, L2 to L1 batch posting
Batch posting continuity measured from finalized L1 data. BS1 nominal · BS2 posting gap detected. Invisible to fee monitors post-EIP-4844.
Three Primitives
Three factual outputs. The agent owns the policy.
Since EIP-4844 (March 2024), L2 sequencer incidents and bridge posting gaps produce no economic signature on L1. Fee monitors are structurally blind to what happens below the surface. Invarians measures it directly, from finalized, public on-chain data.
01
Attestation
The cryptographic envelope. HMAC-SHA256 over the canonical JSON payload, with key identifier and on-chain anchor slot reserved (May 2026). Independently verifiable via
Answers: can I trust this data?
Details →
/v2/verify.Answers: can I trust this data?
02
Regime (SxDx)
The substrate state classification. 12 codes per chain via the structural × demand grid, with signed direction extension (S2+, S2-, D2+, D2-, D2±) on chains with calibrated lower bounds.
Answers: what is the substrate doing now?
Details →
Answers: what is the substrate doing now?
03
Drift Signal
The trend signal. Per-metric
Answers: is it safe to act in the next 30 minutes?
Details →
shift, shift_delta, shift_magnitude_delta over short (~10 h) and long (~30 d) EMAs, plus composite drift per axis.Answers: is it safe to act in the next 30 minutes?
// GET /v2/panel?chains=ethereum,arbitrum&bridges=cctp&include=diagnostic
version: "2.0.0", oracle_status: "OK",
issued_at: "2026-04-30T12:30:00Z",
panel {
l1: [{ chain: "ethereum", regime: "S1D1",
structural: { rhythm: { ratio: 1.0013, shift: 0.0017, shift_magnitude_delta: 0.0008 }, ... },
demand: { tx: { ratio: 1.2037, shift: 0.024, shift_magnitude_delta: 0.018 }, ... },
drift: { structural: 0.0017, demand: 0.04, demand_magnitude_delta: 0.018 }
}],
l2: [{ chain: "arbitrum", regime: "S1D1", ... }],
bridges: [{ id: "arbitrum-ethereum/cctp", type: "cctp", state: "BS1", calibrated: true }]
}
signed_execution_context {
payload_hash: "0x84f5…", signature: "hmac-sha256:…", key_id: "invarians-v1", anchor: null // on-chain May 2026
}
Pattern Reference (the historical frequency of each L1 × L2 × bridge combination) is a complementary research output, exposed via /patterns. It is an analytical reference, not part of the live signed payload.
Coverage snapshot
What you can use today. What is coming.
Usable today (2026-04-29)
- Ethereum, Polygon, signed regimes (12-state) since 2026-04-29
- Solana, Avalanche, 4-state legacy (signed codes scheduled July 2026)
- Arbitrum, Base, Optimism, signed regimes (12-state since v2.0, L2 reaches S2 via
sequencer_publish_latency) - 10 CCTP routes, BS1 / BS2 preliminary calibration on Circle attestation API (since 2026-05-04)
- API v2.0.0, SDK
invarians >= 0.6.0on PyPI, panel endpointGET /v2/panel - HMAC-SHA256 signed attestations, axis-grouped
MetricBlockper observable (ratio,ratio_long,shift) and compositedriftper axis (Primitive 3, Drift Signal)
In panel, calibration pending
- 10 CCIP lanes, captured per message (source-to-execute matched by bytes32
messageId),capability_level: per_message_attested.crypto.anchorisnulltoday; DON multi-sigCommitReportcapture is the next step. - 10 CCTP routes, captured per message with the Circle ECDSA signature (
crypto.anchor: circle_ecdsa), independently verifiable against Circle's published attester public key. BS1 / BS2 classification is preliminary on per-message latency baseline. - Polygon, Avalanche L1, raw capture, MEDIUM and LOW confidence
- Unified
BS1/BS2nomenclature onpanel.bridges[].state,typefield distinguishes protocol - Safety overrides (RMN cursed, Circle API down) fire today, independent of calibration
Coming
- May-June 2026, CCIP cryptographic anchor: DON multi-sig
CommitReportcapture and per-message Merkle inclusion proof (capability_level: per_message_crypto_anchored). - July 2026, Solana full regime calibration
- Q3 2026, ETH ↔ SOL CCIP and CCTP per-message capture once the Solana RPC pipeline is integrated
- Q4 2026, MCP server, first public backtests, third-party fast bridges (Across, Hop)
- 2027, Decentralised INVAR network
Operational roadmap
Where we are. Where we're going.
Live · partial
L1 + L2 Structural Attestation in production
Seven chains live, ETH · POL · SOL · AVA · ARB · BASE · OP. Python SDK, Labs, CRE integration deployed. Calibration status evolving per chain, ETH and POL validated, SOL and AVAX in calibration. Public audit trail on GitHub.
Late April 2026
EVM Full Operational · Centralised
Variable-latency cross-chain surfaces exposed through a single panel: 10 Circle CCTP routes (BS1 / BS2 preliminary) and 10 Chainlink CCIP lanes (raw observability, classification deferred until sustained throughput). Unified
BS1 / BS2 nomenclature, type field distinguishes protocol. Direction agnostic API: GET /v2/panel, the agent composes cross chain routes from L1, L2 and bridge states.Q2–Q3 2026
First CRE Workflow Onchain · MCP Server · Non-EVM Calibration
Execution context bound to certified network state. Agent-native MCP access. Solana and Avalanche structural calibration validated. Third-party bridges (LayerZero, Across) extended. Pattern Reference enters production.
Q4 2026
Trustless Integration, Hybrid Model
Chainlink Functions DON consensus. CCIP L1×L1 · L1×L2 cross-chain execution context. First public backtests.
2027
Protocol Decentralisation
Independent nodes compute. Stake to participate. Computation rewards. Open methodology.