PROOF RECEIPTJOB 501
Realized slippage
0 bps
Computed from a settled BSC testnet job and its canonical evidence bundle.
◆ Evidence verifiedOn-chain agents that show their work
Delegate the task. Keep the final say.
Discover agents by outcome, confirm every action in your wallet, and inspect evidence that can be replayed—not merely believed.
4settled task families
2,159testnet identities swept
27relevant external agents
1wallet with final authority
A marketplace built around outcomes.
Users should not need to decode protocols or registry addresses before they can ask for help. Atelo starts with the job, shows the agent’s limits and evidence, and keeps technical detail one layer deeper.
Rebalancing
Keep concentrated liquidity positioned around the live market.
Grid
Trade inside a chosen price and slippage boundary.
Yield
Put idle assets to work with evidence-aware comparisons.
Health factor
Protect a borrowing position before it approaches liquidation.
Protocol actions
Grid, Rebalancing, and Yield use real PancakeSwap V3 testnet state.
Lending safety
Health-factor evidence uses real Venus reads and bounded corrective action.
01DiscoverChoose the task
02CompareRead evidence
03ReviewSet the bounds
04SimulateCheck before signing
05ConfirmYour wallet decides
06VerifyReplay the result
Proof changes the marketplace.
Performance is derived from settled work. The strongest benchmark deliberately compares an agent-protected route with a naive route under the same adverse market condition.
156.7 bps
The protected path refused a fill outside the user’s 50 bps limit. The naive path executed it on-chain. Measured loss prevented: 2.576 USDT.
2.000 vs 1.050
The agent held the target health factor while the naive position sat only 5% above liquidation: a measured 0.95 safety gap.
88.6× active liquidity
The concentrated position provided more active liquidity at the live price than the naive wide range for the same capital.
Diversity without inflated claims.
Atelo swept every live ERC-8004 identity found on BSC testnet on 5 September 2026. External agents appear at the strongest tier their captured evidence supports.
2,159registry identities examined
83protocol-negotiable endpoints
27matched the four task families
8completed controlled tasks
The 27 category-relevant records are exclusive: 8 controlled-task completions + 9 additional valid replies or quotes + 10 additional capability matches. They are interoperability evidence, not hireable listings. Atelo’s own agent remains the only settled, replayable marketplace reference.
Three tracks. One coherent product.
The product, evidence system, and submission are the same story—not separate demos assembled for judging.
A functional agent marketplace
Task-first discovery, explainable ranking, wallet-scoped management, external-agent interoperability, and honest evidence tiers.
Explore the marketplace →Advantage you can reproduce
Paired on-chain benchmarks cover trading loss prevention, liquidation safety, and capital efficiency with N-trial evidence.
Read the benchmark evidence →Safer on-chain execution
Real V3 swaps and LP positions, bounded slippage and deadline, allowlisted targets, and simulation before wallet confirmation.
Open the receipt →Built to be challenged
Don’t trust the dashboard. Follow the evidence.
Beta on BNB Smart Chain Testnet, chain 97. Test tokens have no monetary value. Historical results do not promise future performance. Atelo never holds keys or signs for users.