Note: This project is open-source and currently under active development. Feel free to explore the codebase, review the implementation, and follow its progress as new features and improvements are added.
Phases 1–12 are implemented, including all 12 active terminal pages. Implementation, local fixture acceptance and live/external-provider acceptance are separate gates. Historical acceptance counts below describe their merge milestones, not the current suite. Phase 6–12 review status is not promoted by local tests; external provider, authority and operational limitations remain tracked in Audit 1.0. Recent hardening closed A1-018 through A1-023; A1-024 through A1-027 are informational scope boundaries documented rather than implemented as new runtime authority. PAPER remains deterministic and crossing-only; TESTNET remains guarded. No mainnet, custody or money movement is supported. History is bounded, in memory and limited to the current session.
HyperAMM converts a mathematical AMM liquidity curve into discrete order-book liquidity for Hyperliquid. It is not an on-chain pool. The system uses virtual constant-product reserves as a deterministic liquidity model, samples that curve around a market-derived fair value, optionally concentrates liquidity near the reference range, normalizes prices/sizes, and reconciles the desired ladder into resting CLOB orders.
Phases 1–12 are implemented / in review, including deterministic supervisory agents, simulation + bounded optimization, research vault/accounting and the full operator terminal as one integrated Python/FastAPI + React/TypeScript system. PAPER is the default execution mode; signed Hyperliquid testnet orders are separately guarded. Mainnet trading, withdrawals, transfers and bridging are deliberately out of scope.
CLOB market making and AMM liquidity have different execution mechanics but can share a useful mathematical representation. HyperAMM treats x*y=k as a quote-generation model instead of a settlement venue:
Hyperliquid L2 / Demo Feed
↓
normalized BBO + freshness
↓
fair value = (bid + ask) / 2
↓
virtual x*y=k reserves
↓
constant / concentrated curve
↓
tick + size discretization
↓
bid/ask QuoteLevel ladder
↓
deterministic risk checks
↓
KEEP / CREATE / REPLACE / CANCEL
↓
PAPER execution or guarded TESTNET
-
Phase 1 — Market data + paper execution: normalized L2/BBO state, WebSocket subscription adapter, initial L2 snapshot, monotonic exchange-time handling, reconnection/degraded states, explicit deterministic demo feed, paper orders/fills, optional testnet adapter.
-
Phase 2 — Virtual constant-product AMM: invariant, marginal price, reserve initialization/recentering, base→quote and quote→base virtual swaps, curve-state calculations.
-
Phase 3 — AMM curve → CLOB compiler: deterministic bid/ask levels, tick and size normalization, stable side ordering, no-cross checks, and stateful quote reconciliation.
-
Phase 4 — Concentrated liquidity: bounded deterministic weighting that shifts more fixed liquidity near fair value as concentration increases.
-
Phase 5 — Inventory-aware quoting: normalized PAPER/TESTNET inventory, bounded reservation-price and side-size skew, hard-limit side suppression, inventory freshness/version checks, and terminal controls/visibility.
-
Phase 6 — Volatility + order-book imbalance adaptation: bounded rolling mid-price volatility, normalized top-N L2 imbalance, widening-only spread adaptation, conservative size/depth reduction, market-state version binding, and terminal explainability.
-
Phase 7 — Perpetual vAMM context: normalized mark/oracle/funding/OI context, bounded perp strategy reference, AMM recentering, shared TESTNET position observability, freshness/version authority, and terminal explainability.
-
Phase 8 — Reference integrity + institutional risk firewall: normalized multi-source price evidence, deterministic quorum/consensus, source health and freshness, signed deviation matrix, projected exposure, liquidation/PnL guards, NORMAL/WIDEN/REDUCE/HALT postures, hysteresis/recovery, SHA-256 authorization fingerprints, and final pre-transmission authority binding.
-
Phase 9 — Supervisory agents (IMPLEMENTED / IN REVIEW): deterministic regime, adverse-selection markout and execution-quality analysis; conservative supervisor composition; bounded quote widening/size/level reduction; agent provenance bound into final authorization.
-
Phase 10 — Strategy simulation + bounded optimization (IMPLEMENTED / IN REVIEW): deterministic scenario/replay datasets, scenario-time PAPER execution, production-stack simulation, research metrics, reproducible fingerprints, and deterministic grid search across an explicit strategy/agent allowlist.
-
Phase 11 — Deterministic research vault/accounting (IMPLEMENTED / IN REVIEW): one shared average-cost authority for runtime/simulation, immutable idempotent ledger, simulated PAPER fees/funding, settled research capital, equity/drawdown, conservative capital reservations, accounting-bound authorization, TESTNET partial truth, read-only APIs and Vault observability.
A minimal Phase 1–4 risk authority enforces freshness, level count, per-order size, aggregate notional, minimum price, quote distance, execution state and a kill switch. The kill switch cancels active strategy orders and blocks new quote generation.
Backend: Python 3.12+, FastAPI, Pydantic v2, Decimal math, asyncio, websockets, httpx, official hyperliquid-python-sdk, pytest. Frontend: React, TypeScript, Vite, TanStack Query, Zustand, Lightweight Charts, native WebSocket.
Copy optional configuration:
cp .env.example .envcd hyperamm/backend
python3.12 -m venv .venv
source .venv/bin/activate
pip install -e '.[test]'
pytest
uvicorn app.main:app --reload --host 127.0.0.1 --port 8000API: http://127.0.0.1:8000 · OpenAPI: http://127.0.0.1:8000/docs
cd hyperamm/frontend
npm install
npm run build
npm run devTerminal: http://127.0.0.1:5173
The provided .env.example deliberately starts with:
MARKET_DATA_MODE=DEMO
EXECUTION_MODE=PAPER
Demo values are generated deterministically and are visibly labeled as simulated. HyperAMM does not silently use demo values when a LIVE feed fails.
To use public Hyperliquid data without a wallet:
MARKET_DATA_MODE=LIVE
PAPER orders are in-memory strategy orders with explicit status, timestamps and client order IDs. The deterministic simulator fills an order only when it touches/crosses the normalized market. Every such fill is labeled SIMULATED PAPER FILL.
Testnet signing is off by default. To enable it intentionally:
EXECUTION_MODE=TESTNET
ENABLE_HYPERLIQUID_TESTNET_ORDERS=true
HYPERLIQUID_ACCOUNT_ADDRESS=0x...
HYPERLIQUID_PRIVATE_KEY=...
The key remains backend-only. The testnet adapter uses post-only Alo limit orders and deterministic SDK client order IDs. No mainnet execution path is implemented.
GET /api/v1/health
GET /api/v1/markets/{market}
GET /api/v1/markets/{market}/book
GET /api/v1/strategy
PUT /api/v1/strategy
POST /api/v1/strategy/start
POST /api/v1/strategy/stop
GET /api/v1/amm/state
GET /api/v1/amm/curve
GET /api/v1/amm/quotes
GET /api/v1/positions
GET /api/v1/vault
GET /api/v1/accounting/pnl
GET /api/v1/accounting/position
GET /api/v1/accounting/ledger
GET /api/v1/accounting/events
GET /api/v1/terminal/history
GET /api/v1/terminal/events
GET /api/v1/market-adaptation
GET /api/v1/perp-context
GET /api/v1/orders
GET /api/v1/fills
GET /api/v1/references
GET /api/v1/risk
GET /api/v1/risk/evidence
GET /api/v1/risk/events
GET /api/v1/risk/authorization
GET /api/v1/agents
GET /api/v1/agents/events
GET /api/v1/simulation/scenarios
POST /api/v1/simulation/run
POST /api/v1/simulation/optimize
POST /api/v1/risk/kill
POST /api/v1/risk/resume
WS /ws/terminal
The terminal uses a dark institutional layout with selected-market/feed/execution/risk state in the header, rolling market + fair-value charting, Hyperliquid L2, an AMM liquidity view, quote ladder, strategy controls, metric cards, kill-switch controls, supervisory-agent observability, and a focused Simulation & Optimization research page. All 12 pages are active: Dashboard, Markets, Strategy, AMM Settings, Execution, Risk, Supervisory Agents, Vault, Analytics, Simulation & Optimization, Logs and Settings; simulation is visibly labeled SIMULATED / NO LIVE ORDERS and has no auto-deploy action.
The backend suite covers constant-product invariants, marginal price, both virtual swap directions, invalid reserves, deterministic sampling, concentrated normalization and concentration behavior, bid/ask ordering, no-cross guarantees, price/size normalization, deterministic quote generation, inventory policy, rolling volatility history, realized-volatility scoring, top-N L2 imbalance, Phase 6 spread/size adaptation, KEEP/CREATE/REPLACE/CANCEL reconciliation, strategy validation, stale feed rejection, risk validation, kill switch, paper lifecycle/fills, API lifecycle, and the disabled-testnet transmission guard.
Live Hyperliquid integration is intentionally not required by normal unit tests.
- PAPER default.
- TESTNET signing requires explicit opt-in plus a backend key.
- No mainnet execution implementation.
- No withdrawal, transfer or bridge methods.
- No browser signing secrets.
.envand generated secrets are excluded from source control/ZIP packaging.- LIVE feed failures are surfaced as degraded/unavailable rather than replaced with fabricated prices.
Phase 8 adds deterministic multi-source reference integrity and institutional risk authorization. Phase 9 supervisory agents are implemented / in review. Phase 10 simulation is offline PAPER research with deterministic crossing-only fills and reuses Phase 11 accounting with optional simulated fees and deterministic funding intervals. It does not model exchange latency, queue priority, hidden liquidity, or stochastic fill probability. Accounting is an in-memory research session ledger; durable wallet custody and production mainnet trading remain out of scope.
- AMM capital semantics (A1-024): virtual reserves and
kdefine curve geometry.total_liquidityis the explicit per-side base-asset quote budget; changingkalone does not imply more quoted capital, TVL or margin. - Scientific-tooling boundary (A1-025): current deterministic strategy/simulation authority does not depend on Pandas or SciPy. Future scientific tooling belongs in optional offline research with immutable exported inputs and proposal-only outputs, never signing/risk/position/capital authority.
- Accounting/custody boundary (A1-026): the vault and ledger are non-custodial, in-memory research accounting. TESTNET economics remain partial where unsupported; deposits, withdrawals, shares and durable production-book recovery are outside current scope.
- Provenance boundary (A1-027): semantic versions and fingerprints identify their declared material decision/input scopes. They are not complete raw-evidence archives, source-revision manifests or automatic Git/dependency provenance.
docs/ARCHITECTURE.mddocs/AMM_MATH.mddocs/HYPERLIQUID_INTEGRATION.mddocs/ROADMAP.mddocs/REFERENCE_INTEGRITY.mddocs/RISK_FIREWALL.mddocs/AGENTS.mddocs/SIMULATION.mddocs/ACCOUNTING.md
Phases 1–8 are implemented. Phase 9 supervisory agents are IMPLEMENTED / IN REVIEW. Phase 10 deterministic simulation + bounded optimization is IMPLEMENTED / IN REVIEW. Phase 11 deterministic research vault/accounting is IMPLEMENTED / IN REVIEW. Phase 12 React operator terminal is IMPLEMENTED / IN REVIEW; see Terminal.
Constant-product incremental reserve movement now drives quote sizes.
Concentrated mode modifies that AMM profile; total_liquidity remains a per-side
base-asset budget including baseline sizes. See AMM math.
Invalid/stale/degraded feeds and generation/risk failures clear the desired ladder and cancel strategy orders. Enabled intent can recover automatically; manual kill requires explicit resume and start. Quote health is shown in the terminal. Unconfirmed cancellation latches HALTED rather than reporting success.
Execution uses a shared lock and final authority checks. TESTNET reconciles tracked orders through WebSocket-triggered and periodic authoritative venue queries, including partial fills and uncertain order states. See architecture and integration. Normal acceptance tests use deterministic fixtures and require no live venue or wallet. Phase 5 is complete; later phases retain the same execution-authority boundaries.
Phase 5 composes after the accepted neutral AMM/concentration ladder and before deterministic risk/reconciliation. PAPER inventory is derived from economic fills only. TESTNET inventory is normalized from Hyperliquid account state and fails closed when authoritative state is missing, stale, malformed, or economically uncertain.
For deviation d = position - target, normal skew uses r = clamp(d / soft_limit, -1, 1). The reservation shift is -r * max_inventory_price_skew_bps; bid and ask sizes use bounded 1 - strength*r and 1 + strength*r side multipliers. At the long hard limit, inventory-increasing bids are omitted; at the short hard limit, inventory-increasing asks are omitted. Existing reconciliation therefore cancels forbidden resting quotes without a second order-management path.
See Inventory Skew for formulas, source semantics, failure behavior, and test coverage.
Phase 6 composes after the accepted Phase 5 inventory policy and before deterministic risk. It uses only normalized Hyperliquid/demo market data already present in the project.
The volatility input is normalized mid price. A bounded rolling window stores unique market observations by exchange sequence/timestamp. Once the configured minimum sample count is reached, per-observation realized volatility is:
r_t = ln(P_t / P_(t-1))
sigma = sqrt(mean(r_t^2))
Before minimum history exists, Phase 6 reports WARMING_UP, exposes no fabricated realized-volatility value, and leaves spread/size adaptation neutral.
Top-N L2 imbalance is base-size depth:
imbalance = (bid_depth - ask_depth) / (bid_depth + ask_depth)
with a bounded range of [-1,+1]. Once volatility is ready, the widening-only spread multiplier is:
1 + volatility_spread_strength * volatility_score
+ imbalance_spread_strength * abs(book_imbalance)
clamped to configured limits. Variable liquidity above the existing Phase 5 base_order_size floor is reduced by bounded global volatility and conservative side-specific imbalance multipliers. Phase 5 hard-limit suppression remains authoritative and Phase 6 cannot restore a removed side.
See Market Adaptation.
Phase 7 keeps raw market fair value separate from a bounded perpetual strategy reference. It normalizes Hyperliquid native markPx, oraclePx, current funding, and openInterest; computes signed basis values and OI notional; blends market/mark/oracle prices with configurable weights; applies a bounded funding shift; clamps the total reference move; then recenters the virtual AMM around that strategy reference before Phase 5 inventory and Phase 6 market adaptation.
PAPER mode requires no wallet and exposes null account-only fields. DEMO perp values are deterministic and explicitly labeled simulated. TESTNET reuses the same user_state() snapshot for Phase 5 inventory and Phase 7 perp-position observability, with an exact signed-position consistency check.
See Perpetual Context.
Phase 8 is the final deterministic authority between strategy proposal and execution. It normalizes RedStone primary-oracle evidence, Hyperliquid native oracle/mark/mid evidence, Kraken exchange BBO evidence, and CoinGecko aggregate evidence into provider-independent contracts. It evaluates source health, per-provider freshness, role-aware quorum, deterministic median consensus, outliers, signed deviations, projected exposure, liquidation distance, fill-derived PAPER PnL, TESTNET equity/drawdown where authoritative values exist, and an explicit NORMAL / WIDEN / REDUCE / HALT state machine.
Only the authorized ladder continues to the existing structural validate_quotes(), reconciliation, and PAPER/guarded TESTNET execution path. Final CREATE/REPLACE authority is bound to market, inventory, perp, reference and risk versions plus SHA-256 quote/evidence/risk fingerprints.
DEMO/PAPER generates deterministic simulated external-reference evidence. LIVE/PAPER can run public reference infrastructure without a trading wallet. Provider credentials remain backend-only.
See Reference Integrity and Risk Firewall.
Phase 9 inserts a bounded, deterministic supervisory layer after Phase 6 strategy adaptation and before the Phase 8 firewall. Regime v2, multi-horizon Toxic-Flow v2, Execution-Quality v2, Liquidity Quality and Perp Crowding consume normalized existing evidence only. Their supervisor can widen spreads, reduce side sizes or trim existing levels; it cannot execute orders, restore suppressed liquidity, clear the manual kill switch or override Phase 8.
FinalQuoteAuthorization binds the material deterministic/heuristic agent version and fingerprint in addition to the existing market, inventory, perp, reference and risk authority. Predictive Adverse Selection is ML SHADOW ONLY, with conditional BID/ASK adverse-fill probabilities and affects_quotes=false; shadow predictions and model changes do not alter quotes or execution authority.
PAPER fill telemetry is explicitly simulated. TESTNET fill-quality metrics remain unavailable where authoritative fill detail does not exist; no slippage or markout is fabricated.
Offline research adds a canonical passive-adverse-v1 feature schema, leakage-aware datasets with independent train/validation identities, optional scikit-learn logistic training (pip install -e '.[test,ml]' from backend), bounded JSON artifacts with SHA-256 provenance, and side/horizon calibration reports. Normal backend installation does not require ML dependencies. The terminal labels heuristic/deterministic/ML SHADOW evidence, and agent APIs publish one coherent cycle snapshot with bounded event filters. No artifact is supplied by default.
Agents recommend. Phase 8 decides. FinalQuoteAuthorization controls execution.
See Supervisory Agents and the agent implementation and ML contracts.
Phase 10 is offline research tooling around the actual HyperAMM strategy stack. It does not contain a second simplified strategy. Each simulation run creates fresh MarketPriceHistory, PAPER execution, Phase 9 telemetry/supervisor, Phase 8 firewall, reconciliation and accounting state, then reuses the existing QuoteEngine → inventory → market adaptation → perp policy → agents → risk → FinalQuoteAuthorization → OrderManager path frame by frame.
Built-in deterministic scenarios cover quiet/trending/mean-reverting/high-volatility markets, book imbalance, flash moves, oracle dislocation, reference degradation, funding stress and liquidity shock. Synthetic references preserve the existing provider identities but are explicitly labeled DEMO / SIMULATED.
The v1 optimizer is deterministic grid search only. It supports an explicit allowlist of strategy/agent research parameters, rejects safety/runtime fields, evaluates an unchanged BASELINE, ranks on training scenarios, evaluates top candidates on separate validation scenarios, and exposes every numeric score component. Results are research rankings under the selected scenarios/objective—not claims of optimality or future profitability.
See Simulation.
The Vault page exposes research settled capital, average-cost position accounting, realized/unrealized PnL, configured PAPER fees and funding, equity/peak/drawdown, simulated full-notional capital reservation and append-only ledger provenance. Runtime PAPER, Phase 8 risk and Phase 10 simulation share one accounting service. Final authorization binds accounting version/fingerprint and stale or changed accounting blocks CREATE/REPLACE while cancellation remains possible.
PAPER accounting is always SIMULATED. TESTNET only exposes existing authoritative account/position evidence and labels full accounting PARTIAL; unavailable cash, realized PnL, fees, funding payments and capital availability remain null. No real custody, deposits, withdrawals, transfers, bridging, investor accounting or fund fees are introduced.
Read-only endpoints: /api/v1/vault, /api/v1/accounting/pnl, /api/v1/accounting/position, /api/v1/accounting/ledger, /api/v1/accounting/events. The terminal WebSocket adds compact accounting summary and recent notices. See Accounting for formulas, defaults, retention, identities and truth boundaries.
Phase 11.1 hardening observes PAPER equity highs after committed fill+fee batches, funding and marks. Vault/API/WebSocket expose identity-based execution/accounting consistency. Missing or conflicting executed fills make capital authority UNAVAILABLE and block CREATE/REPLACE; cancellation remains available. Internal reconciliation books only safe, unfailed pending evidence and never clears a latched error. No public accounting reset/replay endpoint is provided.
The React terminal exposes the existing Phase 1–11.1 pipeline through twelve active pages: Dashboard, Markets, Strategy, AMM Settings, Execution, Risk, Supervisory Agents, Vault, Analytics, Simulation & Optimization, Logs and Settings. It displays retained quote-stage evidence, incremental price/liquidity charts, session analytics, responsive layouts, visible emergency kill and WebSocket diagnostics. Missing or approximate lineage remains qualified by A1-021; chart history ordering (A1-020) and remaining remote/operational limits are tracked in Audit 1.0.
/ws/terminal now carries the code-owned phase12-v1 contract, process-wide
sequence, UTC emission time, pre-agent/post-agent/final quote stages and
observational system health. Read-only /api/v1/terminal/history (3600 retained
points, at most 1000 per response) and /api/v1/terminal/events (500 retained
and at most 500 per response) keep history outside live snapshots.
History is in-memory and current-session only. PAPER/DEMO remains simulated; TESTNET fills and unsupported accounting economics remain explicitly unavailable. Existing risk, supervision, execution and accounting authority/formulas are unchanged. No custody, money movement or autonomous candidate deployment is added. See Terminal architecture, operator pages and limits.
HyperAMM currently supports local, single-operator, research-only use with one backend worker. PAPER is the default; TESTNET still requires explicit mode, order-enable flag and testnet credentials. No mainnet execution is supported. Bind to loopback, keep the frontend local, and do not expose the backend through public/LAN interfaces, reverse proxies, tunnels or port forwarding. Remote/public deployment is unsupported. CORS is browser policy, not authentication.
Startup rejects non-loopback --host/--bind and UVICORN_HOST, and worker
counts other than one in --workers/-w, WEB_CONCURRENCY or UVICORN_WORKERS.
HTTP and WebSocket requests require a loopback peer and strictly parsed local Host.
Browser Origin must match the request origin or an exact local origin configured
in CORS_ORIGINS (default localhost/127.0.0.1/[::1] on port 5173). Unrelated
localhost ports, ambiguous security headers and forwarded headers are rejected.
API write bodies share a 64 KiB pre-parsing bound, including chunked requests. These checks do not
make a loopback proxy or separately launched backend processes supported. Use
one instance only; there is no multi-process coordination. Local Uvicorn reload
remains supported. Starlette's in-process test transport is accepted for tests.
Future remote deployment requires operator authentication/authorization, request/resource admission, WebSocket client controls, TLS/proxy policy, audit logging, and a secrets lifecycle before exposure. These are future work, not features provided by this hardening.
One runtime publisher observes terminal state on a one-second cadence, caches
and serializes the latest phase12-v1 snapshot, and fans it out read-only.
Readers never observe domain state or advance sequence/history/events.
Each client has a one-slot queue: newer snapshots replace pending older ones.
At most 32 terminal clients may subscribe; sends time out after five seconds,
and disconnects cancel tasks and release subscriptions. Sequence gaps can reflect
coalescing, not missing trades. Process/session identity and history/event APIs
retain their existing contract. Before the first publication, readers wait for
the publisher; a domain change appears on its next publication.
Redis is EPHEMERAL DISTRIBUTED INFRASTRUCTURE for terminal Pub/Sub, TTL
snapshot caches, shared bounded WebSocket admission and operational heartbeat.
Optional research coordination adds admission and TTL status/result mirrors around
the existing local PAPER thread worker. REDIS_ENABLED=false is the default;
install the backend [redis] extra to opt in. REDIS_REQUIRED=true fails startup
clearly if Redis is unavailable. Redis-dependent admission fails closed during
outages; local risk, kill/cancel, accounting and execution authority stay in process.
The phase12-v1 contract and local single-operator/single-worker deployment limits
are unchanged. Redis does not enable remote deployment or distributed execution.
PostgreSQL is the future durable operational layer and is not implemented here.
See Redis behavior and configuration and the
repository-specific state inventory/design.
CoinGecko enabled Demo and Pro reference modes require COINGECKO_API_KEY. Only the exact TLS hosts https://api.coingecko.com/api/v3 (Demo / x-cg-demo-api-key) and https://pro-api.coingecko.com/api/v3 (Pro / x-cg-pro-api-key) receive credentials. Missing configuration or untrusted hosts become provider ERROR without a network request. Disabled mode needs no credentials. HTTP 400/401/403 are ERROR; 429, 5xx, network and decoding errors are DEGRADED with bounded retries. CoinGecko remains tertiary and cannot satisfy core quorum.
Optional Yahoo personal research observation: install pip install -e '.[yahoo]'; defaults are YFINANCE_REFERENCE_ENABLED=false, YFINANCE_SYMBOL=ETH-USD, YFINANCE_STALE_AFTER_SECONDS=30. Mapping is explicit: ETH→ETH-USD, BTC→BTC-USD; other startup markets are unavailable, not guessed. The decoded yfinance AsyncWebSocket fields id, price, and Unix-millisecond time are validated against a source-time freshness window. Replays, wrong symbols, invalid values and missing/old/future timestamps are rejected. A bounded reconnect loop cancels listener/watchdog tasks and closes sockets. Missing optional dependency is observational ERROR only.
The separately polled GET /api/v1/references/observations endpoint and Risk-page card expose observational health, source age, prices and available signed comparison bps. Unavailable comparisons remain null. Yahoo is excluded from CORE, ALL, the material reference snapshot, economic version/fingerprint, quorum, outlier decisions, risk, agents, authorization, accounting and execution. It cannot replace RedStone, Kraken or Hyperliquid evidence or authorize trades. The Phase 12 streaming contract remains unchanged.
Licensing and acceptance: Yahoo data is for permitted local personal/research observation only. Do not redistribute it publicly or use it for commercial trading without appropriate rights. Live CoinGecko and Yahoo connections, real schemas and reconnect acceptance remain pending. Phase 8.3.1 fixture and local command results are recorded in the roadmap; they do not establish live-provider acceptance. Historical Audit 2.0 issues A2-001–A2-005 are unchanged. No Actions files, Docker, PostgreSQL, signing changes or new trading authority. No additional oracle providers were introduced; unsupported oracle integrations remain prohibited.
HyperAMM is developed and maintained by:
| Contributor | GitHub | |
|---|---|---|
| Brandon Torres — Creator & Lead Developer | @btorressz | |
| Alexander Delano — Contributor | @thatdela |