FAQ

An independent color-team desk for decision-market governance. Six bounded teams challenge the same record from different sides — evidence, attack, market, playbooks, learning, claims — and every finding stays in its own lane. The job is preventing economic exploits, so the review lands before a market opens, not after it resolves.

01Resolved publishes the common picture: treasuries, proposal records, market math, and published forensics. Chromatic is the adversarial layer that interrogates that picture. It consumes published intelligence, credits it, and does not resell it — and it is not affiliated with, sponsored by, or endorsed by 01Resolved, MetaDAO, or any issuer it covers.

Code auditors read contracts, and post-resolution analytics measure markets that already closed. Neither tells a team what its covenant, window length, and seed depth allow before its first market opens. The published Umbra case is the proof: the market defended the treasury during the window, and the forensics explaining how arrived days after it closed.

Each proposal opens two conditional markets: the pass market prices the token if the proposal executes, the fail market prices the counterfactual, and trades in the losing branch unwind. The verdict is the time-weighted average price over the recording window measured against the DAO's configured threshold — a price someone must hold for days, not touch for a second.

The terms a team commits to at launch: where proceeds go, how spending is authorized, allocation and vesting bands, and what a later reviewer can verify. Chromatic records those terms and checks them tri-state — present, absent, or unknown — and a covenant record is not a proposal, so it never receives an attack-vector classification.

An attacker has to move a time-weighted price and hold it for the rest of the window, while every distorting dollar hands arbitrageurs a paid reason to push back. That defense is priced by depth: the flip-cost scenario states, from explicit assumptions, a lower bound on what moving one conditional pool's print would cost. It is an observation about depth, never a safety rating.

Six bounded lanes with versioned contracts: Pink Team (evidence qualification), Red Team (attack testing), Blue Team (market watch), Yellow Team (playbook building), Purple Team (learning loop), White Team (claim control). Each declares its inputs, outputs, handoffs, and boundary rules in the color-team library, and no team's finding overwrites another's.

The desk's output for one record: six separate findings, one per team, each marked complete, withheld, or standing. Withheld lanes stay visibly withheld. Nothing is folded into a composite.

Four monitored economic attack vectors — each with a shipped detector, named indicators, and the evidence the detector consumes — and two documented vectors described while their detectors are pending. Documented vectors never assess and never classify.

Six versioned checks anchored to the published launch terms — use of proceeds restricted; investor reserve capped and non-dilutable; team allocation in band with milestone vesting; linear unlock schedule; cayman spc/sp entity created; third-party review commissioned — each evaluated present, absent, or unknown, failing closed. Alongside them: the flip-cost scenario at the team's stated depth, and a day-one readiness read of which monitored vectors could run against the team's declared evidence plan.

A count of matched indicators for one monitored vector on one record with complete evidence — HIGH-RISK, ELEVATED, WATCH, or NO-MATCH. It is scoped to that vector. It is never a composite proposal-risk score.

The classification is withheld as UNSCORED, UNKNOWN, STALE, or UNDER_REVIEW, and the missing fields are named. Missing wallet-forensics or market inputs never classify LOW — there is no LOW.

No. NO-MATCH says one vector's indicators did not match on complete evidence. It does not clear the proposal, and the desk never stamps anything safe — a green badge is a request it declines.

Because a single number lets certainty in one lane hide uncertainty in another. Six findings that disagree are the product working.

Current proposal state, DAO thresholds, and pass/fail TWAP accumulators, decoded from confirmed MetaDAO Futarchy v0.6 program accounts. Wallet-forensics lanes stay dark on live rows until an event indexer exists; absent fields are left out, never filled from fixtures.

Every research post carries its citations and a data vintage, and every replay names its source. The registries behind the site ship as saved JSON — the vector library, the team contracts, and this FAQ — regenerated from source and pinned by tests so a page cannot drift from its definitions.

A bounded review, in writing: the covenant record, which monitored vectors can run on day one against the team's declared evidence plan, the flip-cost scenario at stated depth, and the evidence packet a later reviewer would need. The memo is a roster run — the same lanes and the same withhold rules as every public surface.

Green badges and safe stamps; composite scores; reselling 01Resolved's data; code audits; and live classifications the detectors cannot produce from present evidence. The refusals are the contract, not friction against it.