Config Validation

validateConfig checks a policy for domain-specific errors and warnings before deployment. It catches mistakes that TypeScript cannot — like USDC configured with 18 decimals, or an approval threshold below the per-transaction cap.

validateConfig


import { validateConfig } from '@txfence/core'
 
const result = validateConfig(policy)
 
if (!result.valid) {
  console.error('Policy has errors:', result.errors)
  process.exit(1)
}
 
if (result.warnings.length > 0) {
  console.warn('Policy warnings:', result.warnings)
}

Errors

Errors indicate the policy will cause runtime failures. Fix before deploying.

Check	Problem
`chains` is empty	No actions can ever pass
`gasBufferMultiplier < 1.0`	Every simulated transaction fails the gas check
`humanApprovalTimeoutMs < 1000`	Approval windows under 1 second are unusable
`maxSpendPerTx.decimals === 0`	Almost certainly wrong
Mismatched decimals on threshold vs cap	Threshold comparisons will be incorrect

Warnings

Warnings indicate unexpected behavior. Review before deploying.

Check	Problem
USDC/USDT with 18 decimals	Authorizes 1,000,000,000,000× the intended amount
ETH/WBTC with 6 decimals	Decimals mismatch for known tokens
`approvalThreshold < maxSpendPerTx`	Every transaction triggers approval
`gasBufferMultiplier > 3.0`	Unusually high buffer
Empty `allowedContracts` with `requireSimulation: false`	Very permissive policy
Expired `allowedContracts` entries	Will be rejected with `contract_entry_expired`

ConfigValidationResult


interface ConfigValidationResult {
  valid:    boolean
  errors:   ConfigError[]
  warnings: ConfigWarning[]
}
 
interface ConfigWarning {
  field:    string
  message:  string
  severity: 'high' | 'medium' | 'low'
}

CLI integration

txfence check-policy runs validateConfig automatically before evaluating the action. Exits 1 on errors, prints warnings and continues.

MCP tool

txfence_validate_config — AI assistants can validate policies before deploying.

Always run validateConfig in your deployment pipeline. A misconfigured decimal count can authorize orders of magnitude more spend than intended.