- Minimum token amount request (1 token)
- Maximum token amount request (1M tokens)
- Typical request with all parameters (100 tokens, 5 min duration)
- Request with optional fields omitted (no result_ref)
- Request with maximum concurrency (100 concurrent tasks)
- Request with minimum latency (1ms target)
- Insufficient token balance
- Invalid program address
- Malformed parameters
- Expired request
- Successful token transfer to escrow
- Partial refund on completion
- Full refund on failure
- Escrow expiration handling
- Double spend attempt
- Unauthorized escrow release
- Expired escrow release
- Invalid token transfer
- Proposal creation with sufficient stake
- Voting with valid tokens
- Proposal execution on approval
- Vote tally accuracy
- Staking tokens for governance power
- Delegating voting power
- Claiming staking rewards
- Early unstake with penalty
- Proposal with insufficient stake
- Double voting attempt
- Voting after deadline
- Unauthorized proposal execution
- Invalid staking parameters
- Delegation to invalid address
- Early unstake without penalty
- Reward claim before lockup period
- Multiple simultaneous requests
- Race conditions in voting
- Parallel escrow operations
- High volume stress testing
- Failed request retry
- Partial completion handling
- Invalid state recovery
- Network outage scenarios
- Program upgrade verification
- Fee structure modification
- Escrow release authorization
- Governance parameter changes
- Reentrancy attempts
- Arithmetic overflow
- Invalid instruction data
- Unauthorized access attempts
- Model training validation
- Prediction accuracy verification
- Feature extraction testing
- Confidence score calibration
- Model update verification
- Adversarial attack resistance
- Model performance benchmarks
- Edge case handling