Decision workbench

Change the inputs and see why the verdict changes.

This is the sample decision used throughout this project's README (loading…): real output from the same decide() the CLI and the sealing pipeline use, run on synthetic inputs. The three voices, their posts and $TOKENA's market figures are a hand-written test fixture (fixtures/x_three_voices.json, fixtures/market_tokena.json). None of it is a real account, post or price, which is why the page needs no X or RYO call. Changing an input below computes a fresh, clearly labelled hypothetical; the baseline is never altered.

The baseline decision

BASELINE · SAMPLE · SYNTHETIC INPUTSUnchanged, always

loading…

The evidence

Three made-up voices mention $TOKENA, a made-up token. Uncheck one to simulate it never having posted -- the same effect a quiet source or a rate-limited call has on a real sweep.

Change the inputs

Simulate a market field going dark, or move a policy threshold the same way a real operator would tune it. Every one of these is a real field the live engine reads.

Market field unavailable price_usd volume_24h_usd liquidity_usd safety_score

min_independent_voices Policy default: 2

echo_similarity Policy default: 0.6. A repost tagged by the source is caught regardless of this value.

min_evidence_quality Policy default: 0.4

What changed, and which rule caused it · sample inputs, real engine

BASELINE · SAMPLE INPUTS

HYPOTHETICAL · SAMPLE INPUTSnot written anywhere

This branch is computed fresh from the same pure inputs every time and is never appended to a decision ledger. It shows the engine's cause and effect on sample inputs and says nothing about a real market. Real, live decisions are the ones sealed in the repository's runs/ manifests.