TreasureBench

What each strategy can see and do

Each strategy receives only its declared information and capabilities. External adapters are disabled, credential-free, and never executed in CI.

Scroll horizontally to see every column.

Built-in strategy capability contracts
StrategyDefinitionCapabilitiesStatus
blind-distinctCoordinate distinct actions without evidence.11enabled
private-clue-followingAct on each agent's declared private clue.11enabled
consensusPool declared messages into one repeated action.11enabled
pooled-plannerAssign a declared pooled-information portfolio.11enabled
anonymous-marketUse the registered anonymous selected equilibrium.11enabled
posterior-samplingSample actions from each declared posterior.11enabled
ex-ante-role-policyCommit to differentiated contingent roles ex ante.11enabled
sequential-greedyUse visible feedback in registered greedy order.11enabled
marginal-coverage-greedyChoose the largest declared marginal coverage.11enabled
sole-rescue-responseRespond to the registered sole-rescue reward.11enabled
dd006a-mechanismUse the registered weak DD-006A mechanism row.11enabled
dd006b-mechanismUse the registered strict DD-006B mechanism row.11enabled
registered-atlas-architectureExecute one registered coherent Atlas cell.11enabled
private-onlyIgnore the public signal and use private signals only.11enabled
public-onlyAll agents use the public signal.11enabled
designated-readerExactly one designated agent uses the public signal.11enabled
voluntary-attention-equilibriumUse the registered voluntary-attention equilibrium count.11enabled
audience-optimal-assignmentAssign the public signal to the registered optimal audience.11enabled
conditional-private-dominantFollow private evidence on disagreement.11enabled
conditional-public-dominantFollow public evidence on disagreement.11enabled
third-option-contrarianChoose the third label on disagreement.11enabled
threshold-tied-mode-selectionUse the DD-016 registered independent tied-mode selection.11enabled
threshold-minimum-team-plannerAssign minimum teams to the posterior top-L candidates.11enabled
threshold-equilibrium-censusAudit the DD-017 weak-Nash and small-coalition registry.11enabled
dynamic-autonomous-bayesUse the DD-015 full-credit sequential Bayesian policy.11enabled
dynamic-common-information-plannerUse the DD-015 exact common-information planner.11enabled
team-token-mechanismUse the DD-018 fixed balanced team-token rule.11enabled
marginal-team-contributionUse the DD-018 exactly-minimal-team contribution rule.11enabled
universal-pooling-teamUse the DD-018 universal-pooling team response.11enabled