Skip to content

Methodology / AI agent evaluation

Evaluate decisions, not just final answers.

WagerCall is an AI agent evaluation framework for running reproducible, inspectable decisions in versioned game environments. It records what an agent could observe, what it submitted, and what the authoritative environment produced—then leaves scoring and interpretation to the evaluator.

For
Agent developers, evaluators, MCP implementers, and researchers
Outcome
A precise model for using WagerCall evidence without overstating it

Definition

AI agent evaluation examines a trajectory through state and tools

Unlike a one-turn model check, agent evaluation can inspect a sequence of observations, tool calls, actions, errors, recoveries, and outcomes inside an environment.

The final result matters, but it can hide how the agent reached it. Two agents can end with similar point states after making very different decisions; another can choose a sensible action and receive an unfavorable randomized outcome. A useful evaluation preserves enough of the path to separate those stories.

WagerCall provides small, rule-bound environments where the action space, state machine, randomness identity, and public evidence are explicit. That makes a trajectory easier to inspect than an open-ended production task, while remaining only one bounded source of evidence.

Framework scope

WagerCall supplies environments and evidence primitives

What WagerCall v0 provides and what the evaluator provides
WagerCall providesEvaluator provides
Immutable game-version identityThe behavioral question and hypothesis
Validated observations and strict actionsThe agent, model, prompt, and wrapper definitions
Authoritative transitions and event recordsThe repetition schedule and stopping rule
Synthetic-point accounting evidenceDerived measures, statistical analysis, and interpretation
Deterministic environment support for SessionsThreats-to-validity analysis and reporting

Mechanism

The evidence loop has four observable stages

This loop makes tool-use behavior visible. It can show whether an agent grounded its action in current state, produced a legal schema, recovered after a conflict, or repeated a mutation safely. It does not expose private model reasoning.

  1. 01

    Observe

    The agent reads public state plus only the caller-qualified information allowed by the aggregate.

  2. 02

    Decide

    The agent selects one action from a strict, versioned contract.

  3. 03

    Commit

    The server checks identity, expected version, legality, idempotency, and point constraints before transitioning state.

  4. 04

    Inspect

    The evaluator reads the ordered state, attempt, outcome, events, and applicable point evidence.

Critical distinction

Reproducible environments do not make agents deterministic

Once agents choose different actions, later observations may differ. A comparison should identify the branch point and avoid describing every later decision as if it occurred under identical state.

Reproducibility levels in WagerCall
QuestionAccurate answer
Can the environment be rebuilt?Yes, in deterministic Session mode when stable versioned inputs are equal.
Can the canonical transcript be rebuilt?Yes, when the environment and accepted action sequence are also equal.
Will two agents choose the same actions?Not necessarily. Their trajectories can diverge from the same initial environment.
Does a caller-known seed create a blind test?No. Reproducibility and seed secrecy are different properties.

Evaluation protocol

Use a six-part comparison method

  1. 01

    Define one observable question

    Name the tool-use, decision, recovery, or coordination behavior you want to examine.

  2. 02

    Choose the environment

    Match its state, action space, and information boundary to the question.

  3. 03

    Freeze environment variables

    Record version, rules hash, engine build, RNG identity, configuration, seed policy, and trial schedule.

  4. 04

    Document agent variables

    State the model, prompt, policy, wrapper, tool interface, and the one treatment intended to change.

  5. 05

    Run and preserve evidence

    Keep successful, rejected, conflicted, retried, incomplete, and excluded runs.

  6. 06

    Analyze outside v0

    Define derived measures transparently, report uncertainty, and keep conclusions inside the environment's scope.

Interpretation

Outcome-only comparison is incomplete

The ending point state can summarize a run, but trajectory evidence explains which observable decisions, failures, and recoveries contributed to it.

  • Compare legal-action rates separately from game outcomes.
  • Track conflict and retry handling separately from strategic choices.
  • Inspect whether decisions respond coherently to changing public state.
  • Preserve rejected attempts rather than silently dropping them.
  • Treat evaluator-derived measures as analysis, not as native WagerCall scores.

Environment selection

Each game exposes a different decision structure

Performance in a card-game environment does not establish broad intelligence or real-world safety. It provides structured evidence about behavior under the named rules and information boundary.

WagerCall environment fit
EnvironmentDecision structureUseful evidence
Hi-Lo v1Sequential direction and integer stake with public compositionProbability-sensitive decisions across a finite shoe
Blackjack v1Multi-action hand policy against a dealer with private future orderLegal-action use, hand policy, and state-dependent choices
Texas Hold'em v1Independent agents in a persistent multi-hand RoomCoordination, competition, public/private observation use, and action timing

What this cannot prove

State the limits beside the result

  • WagerCall does not reveal hidden chain-of-thought or a complete causal explanation of a model's action.
  • A reproducible run is not automatically blind, independent, or statistically sufficient.
  • Canonical replay is not an execution attestation or regulatory audit certification.
  • Synthetic-point performance is not a monetary result and cannot be redeemed or transferred.
  • One game or run does not prove general model quality, alignment, safety, or superiority.
  • MCP-connected behavior in WagerCall is not proof of general MCP protocol conformance.

Next step

Run a first inspectable workflow

Continue with the WagerCall concepts and action lifecycle.

Run a first inspectable workflow