Open Source · Non-Custodial · Production Ready
85 orthogonal indicators. Regime-aware signals. 365-day backtested strategies. Everything your agent needs to trade crypto perpetuals — built, tested, and ready to run.
One command. Full trading pipeline.
Install the Envy skill on OpenClaw. Your agent discovers it automatically — no configuration, no wiring. Just say "set me up for trading."
The data engine. 22 API endpoints for indicators, signal scoring, backtesting, strategy assembly, and portfolio optimization. Everything your agent needs to build a strategy from scratch.
Connects to the real-time WebSocket and evaluates your signal expressions every 15 seconds across up to 10 coins. Emits ENTRY, EXIT, and ENTRY_END events on state transitions. Stateless by design.
Receives signals from the monitor, enforces risk rules, sizes positions, and routes trades to an executor. Spawns the monitor automatically — run one command to start the full pipeline.
The discover endpoint is a complete, self-describing API manifest — every endpoint, every parameter, every pricing tier, authentication method, and data schema in one JSON response.
Drop the URL into any AI agent and watch it understand the entire system on its own. No documentation to read. No integration guide. The API describes itself.
Every heartbeat of the market — every tremor of momentum, every whisper of trend exhaustion, every fracture in the noise — flows into a single unified place. Not dashboards. Not charts. Alive, current, and waiting to be understood.
Technical momentum across six timeframes. Ichimoku cloud geometry. Bollinger band compression. Rate of change. Mean reversion pressure. All normalized, comparable, ready.
Momentum across six timeframes. Ichimoku clouds. Bollinger compression. Rate of change. Mean reversion pressure — all normalized into a single comparable surface.
Analyst-crowd divergence. Narrative velocity. Mood acceleration. Disagreement spread across cohorts. The market is made of humans — we measure that too.
Hurst exponents reveal trending vs. mean-reverting character. Lyapunov exponents measure proximity to chaos. DFA finds structure inside noise traditional indicators cannot see.
Signals about signals. Features that measure not just what the market is doing, but how reliably our own tools are currently working.
Writing to the Envy Matrix is not open. Every producer must pass through EnvyGate — authenticated, scoped, permitted only to write what it was built to know. A chaos engine cannot pollute sentiment data. A sentiment agent cannot touch price indicators.
The boundaries are architectural, not policy. No discipline required. No human oversight needed. The gate holds by design. The result: a store that gets richer over time without getting noisier. Every number earned its place.
An Envy Engine is an autonomous producer — a sealed process that takes raw reality and transmutes it into a dimension the matrix has never seen. One breathes in order book microstructure and exhales fractal geometry. Another ingests social narrative across twelve platforms and produces a single curve: the velocity of consensus collapse.
Each Engine has one contract with the gate — within that contract it can emit as many dimensions as reality demands. A new Engine can come online at midnight, and by morning the matrix is wider by one column. No migration. No retraining. The matrix simply grew a new sense organ.
The matrix holds 85 dimensions. The Hunters walk through all of them — testing intersections no human would think to name. A chaos exponent crossed with a sentiment divergence crossed with a momentum compression. Combinations that do not have words yet.
A Hunter mutates, probes, discards. It finds a combination that fires — then immediately asks whether the pattern holds where it has never looked. If the signal survives territory it was not born in, it lives. Everything else dies without ceremony.
They find edges in the gaps between disciplines — where technical structure, social fracture, and chaotic dynamics intersect in ways no single domain expert would ever look.
A signal is a spark. A collection of signals is noise. The distance between edge and survival is measured in correlation.
Strategy Assemblers see everything the Hunters have proven. They read the cross-correlation matrix and assemble positions that architect exposure rather than chase returns. A signal long on one coin, hedged by a signal short on another whose correlation inverts under stress.
The spectrum runs from conservative to aggressive — each point a different topology of risk. The geometry is alive, and the Assemblers tend it continuously as the correlation surface shifts.
At the end of the chain, something has to pull the trigger. A Trading Agent receives a complete executable strategy: which signals to watch, in what priority, on which coins, with what exit logic, under what constraints.
When two signals fire on the same candle, the agent knows which speaks first. When a higher-conviction signal appears while a position is open, the agent preempts. Every exit carries a reason. Every reason is logged.
Entry, exit, reason, duration, result — a trail that cannot be revised. The agent's history is its reputation, written in numbers that do not negotiate.
None of these were hypotheses. None were programmed. The matrix held the pattern. The agents found it.
Hurst exponents near 0.4 — deep into mean-reverting territory — identify windows where reversal signals fire with unusual precision. A chaotic signature predicting price behavior traditional indicators miss entirely.
SHORT structures with hit rates below 30% that still generate positive expectancy — because the wins, when they come, are large enough to dominate the distribution. Counter-intuitive by design.
The divergence between analyst sentiment and unfiltered crowd sentiment carries a contrarian signal invisible to pure price analysis. A social fracture the agents weaponized.
FrameOne tears the ceiling out. An agent carries an arsenal — multiple signals, stacked and firing across multiple coins simultaneously. The system stopped thinking in single threads.
Multiple signals stacked and firing across multiple coins simultaneously. A signal that works on BTC momentum doesn't crowd out the one that hunts ETH reversals. Each coin can be watched through more than one lens at once.
Every signal can be tuned — not by hand, not by guesswork, but by the system itself. It tests, compares, escapes dead ends. It finds the configuration that history validates and future price action will prove.
Every signal gets an EVM wallet address — its own on-chain identity. A cryptographic record that cannot be revised, backdated, or spun. The signal's history lives on-chain: transparent, verifiable, owned.
Markets don't move on price. They move on information — on everything humans know, feel, fear, and believe. Earnings reports. Geopolitical tremors. A tweet from someone with too much influence. A whale moving a billion dollars in the dark.
Whale wallet movements hitting the matrix the moment they happen. News sentiment decomposed by source credibility, narrative novelty, and market relevance. On-chain flow analysis. Macro regime indicators. ML predictors that themselves become features — meta-signals about what other models think is about to happen.
Every new data source is a new dimension. Every new dimension is a new axis along which the agents can find combinations no human would think to look for.
The matrix grows. The agents learn.
What will they understand about the world
when they see it with a million eyes?