Non-custodial automation
Users keep ownership while approving bounded autonomous actions.
Product Model
Curated strategies, not raw protocol sprawl.
The product thesis is simple: users should be able to turn a view into a bounded, legible action without becoming protocol operators. Spells achieves that by keeping the mental model stable: discover context, choose a Spell, route it through a Metavault, and monitor the resulting exposure with clear next actions.
Product Pillars
4
Non-custodial automation, opinionated strategies, event-driven execution, persistent risk visibility.
Flagship Spells
3
Everbloom, Ironvine, and Stormforge create a clear risk ladder.
Canonical Position States
3
Active, closed, liquidated. Operational ambiguity is shown separately.
Automation Control
ARM
Autonomy is opt-in, reviewable, and explicitly revocable via DISARM.
Opinionated strategies
Strategies are described by purpose, risk, and expected outcome, not raw calldata steps.
Persistent visibility
What is open, what changed, and what needs attention stay visible after entry.
PRD surfaces
The product is organized around four surfaces that feed each other.
Discovery, execution, monitoring, and automation are not separate products. They form one loop. Discovery creates intent. Execution converts intent into exposure. Monitoring keeps exposure legible. Automation lets the user delegate future action inside limits.
Discovery makes intent legible
Markets, events, and factor summaries should make users confident enough to either create a trigger or review a strategy.
Execution is a deliberate review flow
Users choose a spell, an account, a size, and a bounded action. Estimated fees must be visible before confirmation.
Monitoring keeps state understandable after entry
Users need PnL, risk, actions, alert history, and a human-readable event timeline without reading protocol internals.
Automation stays bounded and reviewable
Triggers remain intentionally constrained: clear conditions, clear action target, explicit state, and revocation that is easy to find.
Strategy semantics
Spells are the product’s primary unit of intention.
A spell is not just a trade. It is a named strategy with purpose, warnings, category, supported context, and an expected user outcome. This is how the product stops raw protocol complexity from becoming the user’s mental model.
Everbloom Yield Loop
Yield-oriented ETH exposure with leverage, basis risk, and borrow-cost sensitivity.
Ironvine
Amplified directional exposure with clearer upside framing and visible liquidation sensitivity.
Stormforge
Highest-conviction, highest-risk spell with very narrow room for error.
Every spell definition must answer
- What is this strategy trying to do?
- Which category and venues does it belong to?
- Which warnings matter before entry?
- What actions will still make sense after entry?
What launch explicitly avoids
- No user-authored strategies.
- No raw protocol builders.
- No strategy marketplace or community publishing.
- No spell presented without risk context.
Account semantics
Metavaults make ownership, isolation, and autonomy understandable.
The account model is central to user trust. Every user gets at least one Metavault, one Metavault is the default, multiple Metavaults can be labeled, and each action clearly shows which account it will affect.
Why Metavaults exist
They separate risk, give the user a stable execution surface, and let the product grant bounded automation without implying asset custody transfer.
What users must always see
Account label, balance context, armed state, fee sufficiency, and where funds move when deposits or withdrawals happen.
Default first account
Users should reach a usable state quickly instead of getting trapped in smart-account setup ceremony.
Purpose-built additional accounts
Users can create and label separate vaults for conservative and higher-risk activity.
Automation remains reversible
ARM must explain limits in plain language. DISARM must remain visible, quick, and hard to misunderstand.
Perps + revenue
GMX expands the product into directional, event-driven exposure.
Perpetuals are the high-risk surface in the system. Launch scope is curated: selected markets, guided long/short setup, leverage and liquidation context, trigger-driven actions, and explicit fee disclosure before confirmation.
Launch scope
Curated perp markets, long/short direction, leverage, collateral choice, and optional exit controls.
User fee framing
Estimated fee before confirmation, cumulative fee visibility in position detail, and balance warnings when fees cannot be covered.
Platform revenue
Open fee at 10 bps, close fee at 5 bps, with a waiver below $100 notional and separate gas-cost recovery.
Perpetuals remain a high-risk surface
Not a full-featured trading terminal
Trigger-compatible at launch
Costs shown before confirmation
Product constraints
The product deliberately says no to several tempting directions.
That restraint is part of the design. Spells is not trying to be a generic trading terminal, a social strategy marketplace, or a scripting environment. It aims to remain understandable to a crypto-literate retail user even when the backend stack grows more sophisticated.
Why generic configurability is out of scope
Unlimited protocol options would break the core product promise. The user
should decide on goals and risk, not manually reconstruct protocol internals.
Why automation needs explicit limits
Automation only works for consumer trust when permissions, account scope,
fee implications, and revocation paths are obvious before the user arms them.
Why the UI stays stable even if infrastructure changes
Users should still understand spells, positions, and alerts if the internal
stack changes later. Product concepts must not expose implementation churn.