Autonomous + deterministic tests for Solana dApps, with instruction-level coverage scored against Anchor IDLs and reproducible runs on mainnet forks.
Solana dApp QA with on-chain awareness
/web3/specs catalogs every Solana program your dApp touches. Each spec is IDL-aware — program ID, detected instructions, verification state per instruction, recommended flows per archetype, and last-scanned timestamp. The spec detail page shows instructions with run counts, expected tx shapes, and the accounts each touches.
Point Autopilot at a connected dApp and it crawls, detects archetypes (swap, staking, NFT mint, governance, …), and generates a covering set of tests automatically. The Web3 overview surfaces archetype-level coverage and instruction-level coverage as separate widgets so you can see where the gaps are.
Every run records every Solana transaction it observed — signature, slot, fee, program ID, instruction name, decoded logs, simulated flag. Issues link back to the specific tx that caused them. Coverage advances on real instructions, not just rendered UI.
/web3/forks manages blockchain fork environments. Pin a fork at a specific slot, seed it with accounts, run the same swap or mint across runs and get identical on-chain state every time. Reproducibility on Web3 stops being aspirational.
Type a prompt like "test the swap flow with a 0.5 SOL input" into the Web3 NL compiler and it produces an Anchor-flavored test definition you can save and run. Useful for stubbing coverage fast or for converting a bug report into a regression test.
/web3/runs adds a cluster chip (devnet / testnet / mainnet-fork), the wallet scenario, and a tx-signature count to every row. Each run detail page exposes healing events — DOM / aria / text / vision-layer selectors the runner auto-recovered, with confidence scores. UI churn on the dApp side stops generating false failures.
/web3/configuration defines wallet scenarios — provider, cluster, starting SOL, preloaded tokens. The runner picks the right scenario at execution time. RPC endpoints are per-cluster, and IDL discovery follows your configured repo refs so program updates flow into specs automatically.
See how web3 fits the rest of Traceback.