Policy profiles decide how KINE handles risk.
Four beta-ready modes tune friction, warnings, and review depth before ALLOW, WARN, or BLOCK.
Threshold story
Profiles tune friction.
Warnings explain consequence.
Red lines override every profile.
Policy profiles
Four modes for different risk appetites. Thresholds change; truth does not.
Conservative
Block aggressively
The strictest profile. It blocks or escalates when there is unclear context, unfamiliar spender behavior, unlimited approvals, or insufficient confidence.
Best for
New users, high-value wallets, treasury wallets, cautious beta testers.
ALLOW examples
- � Known safe read-only actions
- � Low-risk previews
- � Testnet actions with clean simulation
WARN examples
- � New spender
- � Chain mismatch
- � Unusual approval size
BLOCK examples
- � Unlimited approval to unknown spender
- � Known malicious contract
- � Mainnet write path during beta
Balanced
Warn before block
The default KINE posture. It allows clear low-risk actions, warns on ambiguous risk, and blocks serious safety violations.
Best for
Most beta testers and normal wallet safety workflows.
ALLOW examples
- � Expected chain
- � Known app interaction
- � Low-value bounded approval
WARN examples
- � Unverified spender
- � Medium approval exposure
- � Stale session
BLOCK examples
- � High-risk approval
- � Secret request
- � Agent execution authority
Advanced
More user discretion
A less restrictive profile that still explains consequences and blocks red-line violations. It does not remove AION safety boundaries.
Best for
Power users, security researchers, and expert testers who understand wallet risks.
ALLOW examples
- � Expert-confirmed preview
- � Known contract interaction
- � Manually reviewed warning
WARN examples
- � High-value transaction
- � New spender
- � Complex contract path
BLOCK examples
- � Seed phrase request
- � Hidden signing
- � Known malicious target
Testnet Sandbox
Flexible on testnet only
Designed for testnet and simulation. It allows broader experimentation but must never be used as a reason to weaken mainnet safety.
Best for
Testing, demos, QA, and controlled simulation flows.
ALLOW examples
- � Testnet simulation
- � Mock execution flow
- � QA preview scenario
WARN examples
- � Mainnet switch
- � Real asset context
- � Unclear network state
BLOCK examples
- � Mainnet write attempt during beta
- � Secret request
- � Private key handling
Transaction preview model
Every preview should answer these questions before execution.
Actor
Who is requesting the action: user, wallet, dApp, agent, backend, or operator.
Intent
What the action wants to do: approve, transfer, swap, sign, delegate, revoke, or configure.
Target
Which contract, spender, recipient, protocol, or chain is involved.
Asset
Which token, NFT, balance, allowance, or authority is affected.
Consequence
What can happen if the action is accepted, abused, replayed, or misconfigured.
Policy Result
ALLOW, WARN, or BLOCK with reason code, risk level, and receipt.
Policy red lines
These override every profile, including Advanced and Testnet Sandbox.
No policy profile may allow seed phrase entry.
No policy profile may allow hidden signing.
No policy profile may give agents signing authority.
No policy profile may bypass wallet-native consent.
No policy profile may enable uncontrolled mainnet writes during beta.
No policy profile may treat connected as automatically safe.
No policy profile may store private keys.
No policy profile may suppress critical consequences from the user.
How this connects to Preview
The preview page becomes the live testing surface for these policies.
A policy profile tunes the threshold. AION still enforces the red lines.