AxAgentPlaybookEvolveOptions

type AxAgentPlaybookEvolveOptions = object;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L78

Properties

abortSignal?

optional abortSignal: AbortSignal;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L132

apply?

optional apply: boolean;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L129

Keep accepted bullets on the live playbook (default). With false, the playbook is rolled back at the end and the result’s playbookSnapshot carries the accepted state for a later getPlaybook()?.load(...).

epsilon?

optional epsilon: number;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L119

Tolerated held-out drop when accepting a proposal (verify). Default 0.01.

judgeAI?

optional judgeAI: Readonly<AxAIService>;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L99

Scores runs via the built-in judge. Resolution mirrors optimize().

judgeOptions?

optional judgeOptions: AxAgentJudgeOptions;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L100

maxMetricCalls?

optional maxMetricCalls: number;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L110

Budget counting (agent run + judge) pairs across baseline and re-evaluations. Default `max(100, (maxProposals + 1) * (train + validation)

• runsPerTask)`.

maxProposals?

optional maxProposals: number;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L104

Maximum weaknesses mined / proposals evaluated. Default 4.

metric?

optional metric: AxMetricFn;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L102

Optional deterministic scorer replacing the LLM judge.

minHeldInGain?

optional minHeldInGain: number;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L121

Required held-in improvement to accept a proposal (verify). Default 0.05.

onProgress()?

optional onProgress: (event: Readonly<AxAgentPlaybookEvolveProgressEvent>) => void;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L131

Parameters

Returns

void

runsPerTask?

optional runsPerTask: number;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L117

Times each task runs per evaluation, with scores averaged. Default 1. Use 2-3 when the dataset is small: accept/reject compares mean scores, and on a handful of tasks a single lucky or unlucky run can otherwise decide the gate. Each repeat spends budget.

scoreThreshold?

optional scoreThreshold: number;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L123

Records scoring below this count as failures for mining. Default 0.7.

studentAI?

optional studentAI: Readonly<AxAIService>;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L87

Runs the agent during evaluation. Defaults to the agent’s ai.

teacherAI?

optional teacherAI: Readonly<AxAIService>;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L89

Mines weaknesses. Defaults to judgeAI, then the student.

teacherOptions?

optional teacherOptions: AxAIServiceOptions;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L97

AI service options for the weakness miner’s calls. Set useExpensiveModel: 'yes' for a teacher marked isExpensive. The reflection and curation calls that apply each proposal use the playbook’s own teacherOptions (from agent.playbook(...) or the playbook config).

verbose?

optional verbose: boolean;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L130

verify?

optional verify: boolean;

Defined in: https://github.com/ax-llm/ax/blob/3d5aa8dac28a2637e4150f022dcd02e6fc3309f0/src/ax/agent/agentInternal/playbookEvolve/playbookEvolveTypes.ts#L85

Keep only proposals that provably help — re-score train + held-out after each candidate bullet and accept only on a held-in gain without a held-out regression, else roll it back. Default true. With false, mined lessons are applied without the gate (fast trust-batch).