Signal Count

Signal Count turns one market thesis into an auditable risk memo. A coordinator sends the thesis to specialist agents for regime, narrative, and risk analysis, then synthesizes their outputs into scenarios, catalysts, risks, invalidation triggers, and evidence-linked conclusions.

The important part is the proof trail. The app shows which AXL peer handled each role, the dispatch target, specialist output hashes, verifier status, REE receipt status for the risk path, tamper-check evidence for signed execution envelopes, Gensyn Testnet transaction links, reputation/test payout receipts, and indexed chain events.

Signal Count also includes an optional deterministic chain analyst specialist that reads pinned chain-event snapshots instead of producing another LLM sample. It is not a trading bot or price predictor. It is a decision-support proof console for verifying agent work behind a memo.

Signal Count is built with FastAPI, SQLite, Python services, Solidity contracts, Foundry, Docker, and a server-rendered proof-console UI. The coordinator dispatches specialist requests through a local Gensyn AXL mesh using MCP routes shaped like /mcp/{peer_id}/{service_name}. Core specialist services run as separate FastAPI nodes for regime, narrative, and risk analysis, with an optional deterministic chain analyst node backed by pinned chain events.

The risk path can run through Gensyn REE, producing a reproducible execution receipt that is parsed, validated, and displayed in the UI. Signed execution envelopes are checked for tampering by recomputing hashes, recovering signatures, and comparing identity, role, signer, and receipt fields.

Verifier logic hashes specialist outputs and records attestations. Solidity contracts on Gensyn Testnet record task, contribution, and reputation evidence when chain writing is enabled. A local indexer replays confirmed block ranges into SQLite projections so the UI can show chain-backed facts after the run.

Tiny capped native test-ETH payouts are used only as testnet reputation evidence. The full-battle demo script starts the local AXL mesh, specialist nodes, coordinator app, REE path, receipt writing, indexer replay, and leaves a completed browser proof console ready for review.