@xoxoai/checkmate 0.6.7 → 0.7.0

package/dist/runtime/scenario-state.js

@@ -1,128 +0,0 @@
-export class ScenarioState {
-    request;
-    identity;
-    driver;
-    policy;
-    now;
-    startedAtMs;
-    results = [];
-    diagnostics = [];
-    lifecycleFailures = [];
-    cleanupFailure;
-    mutation = 'not-attempted';
-    constructor(options) {
-        this.request = options.request;
-        this.identity = options.identity;
-        this.driver = { ...options.driver };
-        this.policy = { id: options.policy.id, effectiveLimits: { ...options.policy.effectiveLimits } };
-        this.now = options.now ?? Date.now;
-        this.startedAtMs = this.now();
-    }
-    get completedSteps() {
-        return this.results.filter((step) => step.status !== 'not-run');
-    }
-    get hasFailure() {
-        return this.selectFailure() !== undefined;
-    }
-    markMutationPossible() {
-        this.mutation = 'possibly-mutated';
-    }
-    recordStep(result, diagnostics = []) {
-        const expected = this.request.scenario.steps[this.results.length];
-        if (!expected || expected.id !== result.id)
-            throw new Error(`Step '${result.id}' is out of scenario order`);
-        if (this.results.some((step) => step.id === result.id))
-            throw new Error(`Step '${result.id}' was already recorded`);
-        this.results.push(structuredClone(result));
-        this.recordDiagnostics(diagnostics);
-    }
-    blockRemaining(blockedBy) {
-        const blocker = this.results.find((step) => step.id === blockedBy);
-        if (!blocker || blocker.status !== 'failed')
-            throw new Error(`Blocking step '${blockedBy}' is not a failed step`);
-        for (const step of this.request.scenario.steps.slice(this.results.length)) {
-            this.results.push({ id: step.id, status: 'not-run', reason: 'prior-step-failed', blockedBy });
-        }
-    }
-    markNotStarted() {
-        for (const step of this.request.scenario.steps.slice(this.results.length)) {
-            this.results.push({ id: step.id, status: 'not-run', reason: 'scenario-not-started' });
-        }
-    }
-    recordLifecycleFailure(reason, diagnostic) {
-        if (reason === 'interrupted' && this.lifecycleFailures.some((failure) => failure.reason === reason))
-            return;
-        const failure = { reason, ...(diagnostic === undefined ? {} : { diagnostic }) };
-        this.lifecycleFailures.push(failure);
-        if (diagnostic)
-            this.recordDiagnostics([diagnostic]);
-    }
-    latchInterruption(diagnostic) {
-        this.recordLifecycleFailure('interrupted', diagnostic);
-    }
-    recordCleanupFailure(diagnostic) {
-        this.cleanupFailure = { reason: 'driver-teardown-failed', diagnostic };
-        this.recordDiagnostics([diagnostic]);
-    }
-    recordDiagnostics(diagnostics) {
-        for (const diagnostic of diagnostics)
-            this.diagnostics.push({ ...diagnostic });
-    }
-    buildCandidateResult(input) {
-        if (this.results.length !== this.request.scenario.steps.length) {
-            throw new Error('Every declared step must have one terminal scenario state');
-        }
-        const failure = this.selectFailure();
-        const route = failure ? routeForFailure(failure.reason, this.results) : passedRoute();
-        return {
-            kind: 'run-result',
-            schemaVersion: 1,
-            runId: this.identity.runId,
-            scenarioId: this.request.scenario.id,
-            ...route,
-            targetMutation: this.mutation,
-            startedAt: this.identity.startedAt,
-            durationMs: Math.max(0, this.now() - this.startedAtMs),
-            driver: { ...this.driver },
-            policy: { id: this.policy.id, effectiveLimits: { ...this.policy.effectiveLimits } },
-            usage: { ...input.usage },
-            steps: structuredClone(this.results),
-            evidence: {
-                state: input.evidence.state,
-                references: input.evidence.references.map((item) => ({ ...item })),
-            },
-            diagnostics: this.diagnostics.map((diagnostic) => ({ ...diagnostic })),
-        };
-    }
-    selectFailure() {
-        if (this.cleanupFailure)
-            return this.cleanupFailure;
-        const interrupted = this.lifecycleFailures.find((failure) => failure.reason === 'interrupted');
-        if (interrupted)
-            return interrupted;
-        const interruptedStep = this.results.find((step) => step.status === 'failed' && step.reason === 'interrupted');
-        if (interruptedStep)
-            return { reason: 'interrupted' };
-        const evidence = this.lifecycleFailures.find((failure) => failure.reason === 'evidence-write-failed');
-        if (evidence)
-            return evidence;
-        if (this.lifecycleFailures.length > 0)
-            return this.lifecycleFailures[0];
-        const failedStep = this.results.find((step) => step.status === 'failed');
-        if (!failedStep)
-            return undefined;
-        if (failedStep.reason === 'met-expectation')
-            throw new Error('A failed step cannot have met its expectation');
-        return { reason: failedStep.reason };
-    }
-}
-function passedRoute() {
-    return { status: 'passed', category: 'passed', reason: 'scenario-complete' };
-}
-function routeForFailure(reason, steps) {
-    if (reason === 'interrupted')
-        return { status: 'interrupted', category: 'infra', reason };
-    const step = steps.find((candidate) => candidate.status === 'failed' && candidate.reason === reason);
-    return { status: 'failed', category: step?.category ?? 'infra', reason };
-}
-//# sourceMappingURL=scenario-state.js.map

package/dist/runtime/scenario-state.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"scenario-state.js","sourceRoot":"","sources":["../../src/runtime/scenario-state.ts"],"names":[],"mappings":"AA0BA,MAAM,OAAO,aAAa;IACR,OAAO,CAAc;IACrB,QAAQ,CAAa;IACrB,MAAM,CAAoC;IAC1C,MAAM,CAAkD;IACxD,GAAG,CAAc;IACjB,WAAW,CAAQ;IACnB,OAAO,GAAiD,EAAE,CAAA;IAC1D,WAAW,GAAiB,EAAE,CAAA;IAC9B,iBAAiB,GAAc,EAAE,CAAA;IAC1C,cAAc,CAAqB;IACnC,QAAQ,GAAwC,eAAe,CAAA;IAEvE,YAAY,OAA6B;QACxC,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC,OAAO,CAAA;QAC9B,IAAI,CAAC,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAA;QAChC,IAAI,CAAC,MAAM,GAAG,EAAE,GAAG,OAAO,CAAC,MAAM,EAAE,CAAA;QACnC,IAAI,CAAC,MAAM,GAAG,EAAE,EAAE,EAAE,OAAO,CAAC,MAAM,CAAC,EAAE,EAAE,eAAe,EAAE,EAAE,GAAG,OAAO,CAAC,MAAM,CAAC,eAAe,EAAE,EAAE,CAAA;QAC/F,IAAI,CAAC,GAAG,GAAG,OAAO,CAAC,GAAG,IAAI,IAAI,CAAC,GAAG,CAAA;QAClC,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;IAC9B,CAAC;IAED,IAAI,cAAc;QACjB,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,IAAI,EAA8B,EAAE,CAAC,IAAI,CAAC,MAAM,KAAK,SAAS,CAAC,CAAA;IAC5F,CAAC;IAED,IAAI,UAAU;QACb,OAAO,IAAI,CAAC,aAAa,EAAE,KAAK,SAAS,CAAA;IAC1C,CAAC;IAED,oBAAoB;QACnB,IAAI,CAAC,QAAQ,GAAG,kBAAkB,CAAA;IACnC,CAAC;IAED,UAAU,CAAC,MAA0B,EAAE,cAAqC,EAAE;QAC7E,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAA;QACjE,IAAI,CAAC,QAAQ,IAAI,QAAQ,CAAC,EAAE,KAAK,MAAM,CAAC,EAAE;YAAE,MAAM,IAAI,KAAK,CAAC,SAAS,MAAM,CAAC,EAAE,4BAA4B,CAAC,CAAA;QAC3G,IAAI,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,EAAE,KAAK,MAAM,CAAC,EAAE,CAAC;YACrD,MAAM,IAAI,KAAK,CAAC,SAAS,MAAM,CAAC,EAAE,wBAAwB,CAAC,CAAA;QAC5D,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,CAAA;QAC1C,IAAI,CAAC,iBAAiB,CAAC,WAAW,CAAC,CAAA;IACpC,CAAC;IAED,cAAc,CAAC,SAAiB;QAC/B,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,EAAE,KAAK,SAAS,CAAC,CAAA;QAClE,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,MAAM,KAAK,QAAQ;YAC1C,MAAM,IAAI,KAAK,CAAC,kBAAkB,SAAS,wBAAwB,CAAC,CAAA;QACrE,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3E,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,IAAI,CAAC,EAAE,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,mBAAmB,EAAE,SAAS,EAAE,CAAC,CAAA;QAC9F,CAAC;IACF,CAAC;IAED,cAAc;QACb,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;YAC3E,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,IAAI,CAAC,EAAE,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,sBAAsB,EAAE,CAAC,CAAA;QACtF,CAAC;IACF,CAAC;IAED,sBAAsB,CAAC,MAAyB,EAAE,UAAuB;QACxE,IAAI,MAAM,KAAK,aAAa,IAAI,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,KAAK,MAAM,CAAC;YAAE,OAAM;QAC3G,MAAM,OAAO,GAAG,EAAE,MAAM,EAAE,GAAG,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,CAAC,EAAE,CAAA;QAC/E,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;QACpC,IAAI,UAAU;YAAE,IAAI,CAAC,iBAAiB,CAAC,CAAC,UAAU,CAAC,CAAC,CAAA;IACrD,CAAC;IAED,iBAAiB,CAAC,UAAuB;QACxC,IAAI,CAAC,sBAAsB,CAAC,aAAa,EAAE,UAAU,CAAC,CAAA;IACvD,CAAC;IAED,oBAAoB,CAAC,UAAsB;QAC1C,IAAI,CAAC,cAAc,GAAG,EAAE,MAAM,EAAE,wBAAwB,EAAE,UAAU,EAAE,CAAA;QACtE,IAAI,CAAC,iBAAiB,CAAC,CAAC,UAAU,CAAC,CAAC,CAAA;IACrC,CAAC;IAED,iBAAiB,CAAC,WAAkC;QACnD,KAAK,MAAM,UAAU,IAAI,WAAW;YAAE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,EAAE,GAAG,UAAU,EAAE,CAAC,CAAA;IAC/E,CAAC;IAED,oBAAoB,CAAC,KAGpB;QACA,IAAI,IAAI,CAAC,OAAO,CAAC,MAAM,KAAK,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;YAChE,MAAM,IAAI,KAAK,CAAC,2DAA2D,CAAC,CAAA;QAC7E,CAAC;QACD,MAAM,OAAO,GAAG,IAAI,CAAC,aAAa,EAAE,CAAA;QACpC,MAAM,KAAK,GAAG,OAAO,CAAC,CAAC,CAAC,eAAe,CAAC,OAAO,CAAC,MAAM,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAA;QACrF,OAAO;YACN,IAAI,EAAE,YAAY;YAClB,aAAa,EAAE,CAAC;YAChB,KAAK,EAAE,IAAI,CAAC,QAAQ,CAAC,KAAK;YAC1B,UAAU,EAAE,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE;YACpC,GAAG,KAAK;YACR,cAAc,EAAE,IAAI,CAAC,QAAQ;YAC7B,SAAS,EAAE,IAAI,CAAC,QAAQ,CAAC,SAAS;YAClC,UAAU,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,IAAI,CAAC,WAAW,CAAC;YACtD,MAAM,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,EAAE;YAC1B,MAAM,EAAE,EAAE,EAAE,EAAE,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,eAAe,EAAE,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,eAAe,EAAE,EAAE;YACnF,KAAK,EAAE,EAAE,GAAG,KAAK,CAAC,KAAK,EAAE;YACzB,KAAK,EAAE,eAAe,CAAC,IAAI,CAAC,OAAO,CAAC;YACpC,QAAQ,EAAE;gBACT,KAAK,EAAE,KAAK,CAAC,QAAQ,CAAC,KAAK;gBAC3B,UAAU,EAAE,KAAK,CAAC,QAAQ,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,CAAC,CAAC;aAClE;YACD,WAAW,EAAE,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,UAAU,EAAE,CAAC,CAAC;SACtE,CAAA;IACF,CAAC;IAEO,aAAa;QACpB,IAAI,IAAI,CAAC,cAAc;YAAE,OAAO,IAAI,CAAC,cAAc,CAAA;QACnD,MAAM,WAAW,GAAG,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,KAAK,aAAa,CAAC,CAAA;QAC9F,IAAI,WAAW;YAAE,OAAO,WAAW,CAAA;QACnC,MAAM,eAAe,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CACxC,CAAC,IAAI,EAA8B,EAAE,CAAC,IAAI,CAAC,MAAM,KAAK,QAAQ,IAAI,IAAI,CAAC,MAAM,KAAK,aAAa,CAC/F,CAAA;QACD,IAAI,eAAe;YAAE,OAAO,EAAE,MAAM,EAAE,aAAa,EAAE,CAAA;QACrD,MAAM,QAAQ,GAAG,IAAI,CAAC,iBAAiB,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,KAAK,uBAAuB,CAAC,CAAA;QACrG,IAAI,QAAQ;YAAE,OAAO,QAAQ,CAAA;QAC7B,IAAI,IAAI,CAAC,iBAAiB,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,IAAI,CAAC,iBAAiB,CAAC,CAAC,CAAC,CAAA;QACvE,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,IAAI,EAA8B,EAAE,CAAC,IAAI,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAA;QACpG,IAAI,CAAC,UAAU;YAAE,OAAO,SAAS,CAAA;QACjC,IAAI,UAAU,CAAC,MAAM,KAAK,iBAAiB;YAAE,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;QAC7G,OAAO,EAAE,MAAM,EAAE,UAAU,CAAC,MAAM,EAAE,CAAA;IACrC,CAAC;CACD;AAED,SAAS,WAAW;IACnB,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,mBAAmB,EAAE,CAAA;AAC7E,CAAC;AAED,SAAS,eAAe,CACvB,MAA+C,EAC/C,KAAyD;IAEzD,IAAI,MAAM,KAAK,aAAa;QAAE,OAAO,EAAE,MAAM,EAAE,aAAa,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,CAAA;IACzF,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CACtB,CAAC,SAAS,EAAmC,EAAE,CAAC,SAAS,CAAC,MAAM,KAAK,QAAQ,IAAI,SAAS,CAAC,MAAM,KAAK,MAAM,CAC5G,CAAA;IACD,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,IAAI,OAAO,EAAE,MAAM,EAAE,CAAA;AACzE,CAAC"}

package/dist/runtime/step-execution.d.ts

@@ -1,27 +0,0 @@
-import type { AiClient } from '../ai/client.js';
-import type { RuntimeConfig } from './config.js';
-import type { DriverSession, StepIntent } from '../driver.js';
-import type { RuntimeLogger } from '../logging/types.js';
-import { DiagnosticSanitizer } from '../redaction/diagnostic-sanitizer.js';
-import type { ToolRegistry } from '../tools/registry.js';
-import type { StepControl } from './scenario-control.js';
-import type { InternalStepReport } from './types.js';
-import { ScenarioUsageTracker } from './usage-tracker.js';
-export type StepExecutionDependencies = {
-    config: RuntimeConfig;
-    aiClient: AiClient;
-    toolRegistry: ToolRegistry;
-    driverSession: DriverSession;
-    usageTracker: ScenarioUsageTracker;
-    driverId: string;
-    redact: boolean;
-    diagnosticSanitizer: DiagnosticSanitizer;
-    logger: RuntimeLogger;
-};
-/** Authoritative driver-backed model/tool loop. */
-export declare class StepExecution {
-    private readonly dependencies;
-    constructor(dependencies: StepExecutionDependencies);
-    run(step: StepIntent, control: StepControl): Promise<InternalStepReport>;
-}
-//# sourceMappingURL=step-execution.d.ts.map

package/dist/runtime/step-execution.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"step-execution.d.ts","sourceRoot":"","sources":["../../src/runtime/step-execution.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,iBAAiB,CAAA;AAI/C,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,aAAa,CAAA;AAChD,OAAO,KAAK,EAAE,aAAa,EAAE,UAAU,EAAE,MAAM,cAAc,CAAA;AAC7D,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAA;AACxD,OAAO,EAAE,mBAAmB,EAAE,MAAM,sCAAsC,CAAA;AAG1E,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAA;AAIxD,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,uBAAuB,CAAA;AACxD,OAAO,KAAK,EAAE,kBAAkB,EAA6B,MAAM,YAAY,CAAA;AAC/E,OAAO,EAGN,oBAAoB,EAEpB,MAAM,oBAAoB,CAAA;AAE3B,MAAM,MAAM,yBAAyB,GAAG;IACvC,MAAM,EAAE,aAAa,CAAA;IACrB,QAAQ,EAAE,QAAQ,CAAA;IAClB,YAAY,EAAE,YAAY,CAAA;IAC1B,aAAa,EAAE,aAAa,CAAA;IAC5B,YAAY,EAAE,oBAAoB,CAAA;IAClC,QAAQ,EAAE,MAAM,CAAA;IAChB,MAAM,EAAE,OAAO,CAAA;IACf,mBAAmB,EAAE,mBAAmB,CAAA;IACxC,MAAM,EAAE,aAAa,CAAA;CACrB,CAAA;AAED,mDAAmD;AACnD,qBAAa,aAAa;IACb,OAAO,CAAC,QAAQ,CAAC,YAAY;gBAAZ,YAAY,EAAE,yBAAyB;IAE9D,GAAG,CAAC,IAAI,EAAE,UAAU,EAAE,OAAO,EAAE,WAAW,GAAG,OAAO,CAAC,kBAAkB,CAAC;CAoI9E"}

package/dist/runtime/step-execution.js

@@ -1,143 +0,0 @@
-import { MessageHistory } from '../ai/message-history.js';
-import { STEP_START_USER_PROMPT, STEP_SYSTEM_PROMPT } from '../ai/prompts.js';
-import { TurnProcessor } from '../ai/turn-processor.js';
-import { ToolDispatchError } from '../tools/dispatcher.js';
-import { LoopDetector } from '../tools/loop-detector.js';
-import { DriverBoundaryError } from './driver-boundary.js';
-import { buildInitialDriverContext, buildPostToolDriverContext, isEphemeralDriverMessage } from './driver-session.js';
-import { InternalStepEvidence } from './internal-step-evidence.js';
-import { ProviderUsageInvalidError, ProviderUsageUnavailableError, TokenBudgetExceededError, } from './usage-tracker.js';
-/** Authoritative driver-backed model/tool loop. */
-export class StepExecution {
-    dependencies;
-    constructor(dependencies) {
-        this.dependencies = dependencies;
-    }
-    async run(step, control) {
-        const { config, aiClient, toolRegistry, driverSession, usageTracker, driverId, redact, diagnosticSanitizer, logger, } = this.dependencies;
-        const evidence = new InternalStepEvidence(step, driverId, redact, diagnosticSanitizer, control.now);
-        const checkpoint = usageTracker.beginStep();
-        const messages = [];
-        const ephemeralMessages = new Set();
-        const turnProcessor = new TurnProcessor({
-            config,
-            toolRegistry,
-            loopDetector: new LoopDetector(config.loopMaxRepetitions),
-            evidence,
-            logger,
-        });
-        messages.push(...new MessageHistory().buildInitialMessages({
-            systemPrompt: STEP_SYSTEM_PROMPT([...driverSession.instructions]),
-            userPrompt: STEP_START_USER_PROMPT(step),
-        }));
-        logger.info(`step started: ${step.id}`);
-        let turns = 0;
-        try {
-            try {
-                appendDriverContext(messages, ephemeralMessages, await buildInitialDriverContext(driverSession, step, control));
-            }
-            catch (error) {
-                return finish(evidence, usageTracker, checkpoint, failureFrom(error, control, turns, diagnosticSanitizer), logger);
-            }
-            for (;;) {
-                if (turns >= config.turnCap) {
-                    return finish(evidence, usageTracker, checkpoint, { outcome: 'failed', reason: 'turn-cap-exceeded', turns }, logger);
-                }
-                const expired = control.poll();
-                if (expired.expired) {
-                    return finish(evidence, usageTracker, checkpoint, { outcome: 'failed', reason: expired.reason, turns }, logger);
-                }
-                turns++;
-                try {
-                    const { response, assistantMessages } = await aiClient.send(messages, {
-                        step,
-                        signal: control.signal,
-                    });
-                    usageTracker.record(response.usage);
-                    assertControlLive(control, 'provider-response');
-                    messages.push(...assistantMessages);
-                    const outcome = await turnProcessor.process({ response, step, turn: turns, control });
-                    assertControlLive(control, 'turn-processing');
-                    if (outcome.kind === 'assertion') {
-                        return finish(evidence, usageTracker, checkpoint, {
-                            outcome: outcome.passed ? 'passed' : 'failed',
-                            reason: outcome.passed ? 'met-expectation' : 'failed-expectation',
-                            actual: outcome.actual,
-                            turns,
-                        }, logger);
-                    }
-                    if (outcome.kind === 'stuck') {
-                        return finish(evidence, usageTracker, checkpoint, {
-                            outcome: 'failed',
-                            reason: 'loop-detected',
-                            actual: 'the model repeated the same tool calls without reaching a result',
-                            turns,
-                        }, logger);
-                    }
-                    dropEphemeralMessages(messages, ephemeralMessages);
-                    messages.push(...outcome.messages);
-                    appendDriverContext(messages, ephemeralMessages, await buildPostToolDriverContext(driverSession, step, turns, outcome.toolResults, control));
-                }
-                catch (error) {
-                    return finish(evidence, usageTracker, checkpoint, failureFrom(error, control, turns, diagnosticSanitizer), logger);
-                }
-            }
-        }
-        finally {
-            control.dispose();
-        }
-    }
-}
-function finish(evidence, usageTracker, checkpoint, termination, logger) {
-    if (termination.reason === 'step-timeout' || termination.reason === 'scenario-timeout') {
-        evidence.recordDiagnostic({
-            code: 'expired-boundary',
-            path: '',
-            message: termination.actual ?? `Step execution expired with ${termination.reason}`,
-        });
-    }
-    const report = evidence.buildReport(termination, usageTracker.stepUsage(checkpoint));
-    logger.info(`step finished: ${report.step.id} ${report.outcome} (${report.category} / ${report.reason})`);
-    return report;
-}
-function appendDriverContext(messages, ephemeralMessages, context) {
-    for (const message of context) {
-        messages.push(message);
-        if (isEphemeralDriverMessage(message))
-            ephemeralMessages.add(message);
-    }
-}
-function dropEphemeralMessages(messages, ephemeralMessages) {
-    if (ephemeralMessages.size === 0)
-        return;
-    const retained = messages.filter((message) => !ephemeralMessages.has(message));
-    messages.length = 0;
-    messages.push(...retained);
-    ephemeralMessages.clear();
-}
-function assertControlLive(control, operation) {
-    const expired = control.poll();
-    if (expired.expired)
-        throw new DriverBoundaryError(operation, expired.reason);
-}
-function failureFrom(error, control, turns, sanitizer) {
-    const actual = sanitizer.error(error);
-    const expired = control.poll();
-    if (expired.expired)
-        return { outcome: 'failed', reason: expired.reason, actual, turns };
-    if (error instanceof DriverBoundaryError)
-        return { outcome: 'failed', reason: error.reason, actual, turns };
-    if (error instanceof TokenBudgetExceededError) {
-        return { outcome: 'failed', reason: 'token-budget-exceeded', actual, turns };
-    }
-    if (error instanceof ProviderUsageUnavailableError || error instanceof ProviderUsageInvalidError) {
-        return { outcome: 'failed', reason: 'provider-error', actual, turns };
-    }
-    if (error instanceof ToolDispatchError)
-        return { outcome: 'failed', reason: 'tool-error', actual, turns };
-    return { outcome: 'failed', reason: driverInfraReason(error), actual, turns };
-}
-function driverInfraReason(error) {
-    return error instanceof ToolDispatchError ? 'tool-error' : 'provider-error';
-}
-//# sourceMappingURL=step-execution.js.map

package/dist/runtime/step-execution.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"step-execution.js","sourceRoot":"","sources":["../../src/runtime/step-execution.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AACzD,OAAO,EAAE,sBAAsB,EAAE,kBAAkB,EAAE,MAAM,kBAAkB,CAAA;AAC7E,OAAO,EAAE,aAAa,EAAE,MAAM,yBAAyB,CAAA;AAKvD,OAAO,EAAE,iBAAiB,EAAE,MAAM,wBAAwB,CAAA;AAC1D,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AAExD,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAC1D,OAAO,EAAE,yBAAyB,EAAE,0BAA0B,EAAE,wBAAwB,EAAE,MAAM,qBAAqB,CAAA;AACrH,OAAO,EAAE,oBAAoB,EAAgC,MAAM,6BAA6B,CAAA;AAGhG,OAAO,EACN,yBAAyB,EACzB,6BAA6B,EAE7B,wBAAwB,GACxB,MAAM,oBAAoB,CAAA;AAc3B,mDAAmD;AACnD,MAAM,OAAO,aAAa;IACI;IAA7B,YAA6B,YAAuC;QAAvC,iBAAY,GAAZ,YAAY,CAA2B;IAAG,CAAC;IAExE,KAAK,CAAC,GAAG,CAAC,IAAgB,EAAE,OAAoB;QAC/C,MAAM,EACL,MAAM,EACN,QAAQ,EACR,YAAY,EACZ,aAAa,EACb,YAAY,EACZ,QAAQ,EACR,MAAM,EACN,mBAAmB,EACnB,MAAM,GACN,GAAG,IAAI,CAAC,YAAY,CAAA;QACrB,MAAM,QAAQ,GAAG,IAAI,oBAAoB,CAAC,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,mBAAmB,EAAE,OAAO,CAAC,GAAG,CAAC,CAAA;QACnG,MAAM,UAAU,GAAG,YAAY,CAAC,SAAS,EAAE,CAAA;QAC3C,MAAM,QAAQ,GAAiC,EAAE,CAAA;QACjD,MAAM,iBAAiB,GAAG,IAAI,GAAG,EAA8B,CAAA;QAC/D,MAAM,aAAa,GAAG,IAAI,aAAa,CAAC;YACvC,MAAM;YACN,YAAY;YACZ,YAAY,EAAE,IAAI,YAAY,CAAC,MAAM,CAAC,kBAAkB,CAAC;YACzD,QAAQ;YACR,MAAM;SACN,CAAC,CAAA;QACF,QAAQ,CAAC,IAAI,CACZ,GAAG,IAAI,cAAc,EAAE,CAAC,oBAAoB,CAAC;YAC5C,YAAY,EAAE,kBAAkB,CAAC,CAAC,GAAG,aAAa,CAAC,YAAY,CAAC,CAAC;YACjE,UAAU,EAAE,sBAAsB,CAAC,IAAI,CAAC;SACxC,CAAC,CACF,CAAA;QACD,MAAM,CAAC,IAAI,CAAC,iBAAiB,IAAI,CAAC,EAAE,EAAE,CAAC,CAAA;QAEvC,IAAI,KAAK,GAAG,CAAC,CAAA;QACb,IAAI,CAAC;YACJ,IAAI,CAAC;gBACJ,mBAAmB,CAClB,QAAQ,EACR,iBAAiB,EACjB,MAAM,yBAAyB,CAAC,aAAa,EAAE,IAAI,EAAE,OAAO,CAAC,CAC7D,CAAA;YACF,CAAC;YAAC,OAAO,KAAK,EAAE,CAAC;gBAChB,OAAO,MAAM,CACZ,QAAQ,EACR,YAAY,EACZ,UAAU,EACV,WAAW,CAAC,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,mBAAmB,CAAC,EACvD,MAAM,CACN,CAAA;YACF,CAAC;YAED,SAAS,CAAC;gBACT,IAAI,KAAK,IAAI,MAAM,CAAC,OAAO,EAAE,CAAC;oBAC7B,OAAO,MAAM,CACZ,QAAQ,EACR,YAAY,EACZ,UAAU,EACV,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,mBAAmB,EAAE,KAAK,EAAE,EACzD,MAAM,CACN,CAAA;gBACF,CAAC;gBACD,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;gBAC9B,IAAI,OAAO,CAAC,OAAO,EAAE,CAAC;oBACrB,OAAO,MAAM,CACZ,QAAQ,EACR,YAAY,EACZ,UAAU,EACV,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,KAAK,EAAE,EACpD,MAAM,CACN,CAAA;gBACF,CAAC;gBAED,KAAK,EAAE,CAAA;gBACP,IAAI,CAAC;oBACJ,MAAM,EAAE,QAAQ,EAAE,iBAAiB,EAAE,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE;wBACrE,IAAI;wBACJ,MAAM,EAAE,OAAO,CAAC,MAAM;qBACtB,CAAC,CAAA;oBACF,YAAY,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAA;oBACnC,iBAAiB,CAAC,OAAO,EAAE,mBAAmB,CAAC,CAAA;oBAC/C,QAAQ,CAAC,IAAI,CAAC,GAAG,iBAAiB,CAAC,CAAA;oBACnC,MAAM,OAAO,GAAG,MAAM,aAAa,CAAC,OAAO,CAAC,EAAE,QAAQ,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,CAAA;oBACrF,iBAAiB,CAAC,OAAO,EAAE,iBAAiB,CAAC,CAAA;oBAE7C,IAAI,OAAO,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;wBAClC,OAAO,MAAM,CACZ,QAAQ,EACR,YAAY,EACZ,UAAU,EACV;4BACC,OAAO,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,QAAQ;4BAC7C,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,oBAAoB;4BACjE,MAAM,EAAE,OAAO,CAAC,MAAM;4BACtB,KAAK;yBACL,EACD,MAAM,CACN,CAAA;oBACF,CAAC;oBACD,IAAI,OAAO,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;wBAC9B,OAAO,MAAM,CACZ,QAAQ,EACR,YAAY,EACZ,UAAU,EACV;4BACC,OAAO,EAAE,QAAQ;4BACjB,MAAM,EAAE,eAAe;4BACvB,MAAM,EAAE,kEAAkE;4BAC1E,KAAK;yBACL,EACD,MAAM,CACN,CAAA;oBACF,CAAC;oBAED,qBAAqB,CAAC,QAAQ,EAAE,iBAAiB,CAAC,CAAA;oBAClD,QAAQ,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAA;oBAClC,mBAAmB,CAClB,QAAQ,EACR,iBAAiB,EACjB,MAAM,0BAA0B,CAAC,aAAa,EAAE,IAAI,EAAE,KAAK,EAAE,OAAO,CAAC,WAAW,EAAE,OAAO,CAAC,CAC1F,CAAA;gBACF,CAAC;gBAAC,OAAO,KAAK,EAAE,CAAC;oBAChB,OAAO,MAAM,CACZ,QAAQ,EACR,YAAY,EACZ,UAAU,EACV,WAAW,CAAC,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,mBAAmB,CAAC,EACvD,MAAM,CACN,CAAA;gBACF,CAAC;YACF,CAAC;QACF,CAAC;gBAAS,CAAC;YACV,OAAO,CAAC,OAAO,EAAE,CAAA;QAClB,CAAC;IACF,CAAC;CACD;AAED,SAAS,MAAM,CACd,QAA8B,EAC9B,YAAkC,EAClC,UAAyD,EACzD,WAAoC,EACpC,MAAqB;IAErB,IAAI,WAAW,CAAC,MAAM,KAAK,cAAc,IAAI,WAAW,CAAC,MAAM,KAAK,kBAAkB,EAAE,CAAC;QACxF,QAAQ,CAAC,gBAAgB,CAAC;YACzB,IAAI,EAAE,kBAAkB;YACxB,IAAI,EAAE,EAAE;YACR,OAAO,EAAE,WAAW,CAAC,MAAM,IAAI,+BAA+B,WAAW,CAAC,MAAM,EAAE;SAClF,CAAC,CAAA;IACH,CAAC;IACD,MAAM,MAAM,GAAG,QAAQ,CAAC,WAAW,CAAC,WAAW,EAAE,YAAY,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC,CAAA;IACpF,MAAM,CAAC,IAAI,CAAC,kBAAkB,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,MAAM,CAAC,OAAO,KAAK,MAAM,CAAC,QAAQ,MAAM,MAAM,CAAC,MAAM,GAAG,CAAC,CAAA;IACzG,OAAO,MAAM,CAAA;AACd,CAAC;AAED,SAAS,mBAAmB,CAC3B,QAAsC,EACtC,iBAAkD,EAClD,OAAqC;IAErC,KAAK,MAAM,OAAO,IAAI,OAAO,EAAE,CAAC;QAC/B,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;QACtB,IAAI,wBAAwB,CAAC,OAAO,CAAC;YAAE,iBAAiB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IACtE,CAAC;AACF,CAAC;AAED,SAAS,qBAAqB,CAC7B,QAAsC,EACtC,iBAAkD;IAElD,IAAI,iBAAiB,CAAC,IAAI,KAAK,CAAC;QAAE,OAAM;IACxC,MAAM,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,iBAAiB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAA;IAC9E,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAA;IACnB,QAAQ,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,CAAA;IAC1B,iBAAiB,CAAC,KAAK,EAAE,CAAA;AAC1B,CAAC;AAED,SAAS,iBAAiB,CAAC,OAAoB,EAAE,SAAiB;IACjE,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;IAC9B,IAAI,OAAO,CAAC,OAAO;QAAE,MAAM,IAAI,mBAAmB,CAAC,SAAS,EAAE,OAAO,CAAC,MAAM,CAAC,CAAA;AAC9E,CAAC;AAED,SAAS,WAAW,CACnB,KAAc,EACd,OAAoB,EACpB,KAAa,EACb,SAA8B;IAE9B,MAAM,MAAM,GAAG,SAAS,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;IACrC,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;IAC9B,IAAI,OAAO,CAAC,OAAO;QAAE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,CAAA;IACxF,IAAI,KAAK,YAAY,mBAAmB;QAAE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,CAAA;IAC3G,IAAI,KAAK,YAAY,wBAAwB,EAAE,CAAC;QAC/C,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,uBAAuB,EAAE,MAAM,EAAE,KAAK,EAAE,CAAA;IAC7E,CAAC;IACD,IAAI,KAAK,YAAY,6BAA6B,IAAI,KAAK,YAAY,yBAAyB,EAAE,CAAC;QAClG,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,gBAAgB,EAAE,MAAM,EAAE,KAAK,EAAE,CAAA;IACtE,CAAC;IACD,IAAI,KAAK,YAAY,iBAAiB;QAAE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,KAAK,EAAE,CAAA;IACzG,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,iBAAiB,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAA;AAC9E,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAc;IACxC,OAAO,KAAK,YAAY,iBAAiB,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,gBAAgB,CAAA;AAC5E,CAAC"}

package/dist/runtime/text.d.ts

@@ -1,10 +0,0 @@
-/**
- * Removes the common leading indentation from a multi-line string.
- *
- * Steps are written as template literals in a spec file, so `action` and `expect` arrive
- * carrying whatever indentation the surrounding code happened to have. That indentation is
- * an artifact of the source layout, not of the step, and it should not reach the report or
- * the failure message.
- */
-export declare function dedent(value: string): string;
-//# sourceMappingURL=text.d.ts.map

package/dist/runtime/text.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"text.d.ts","sourceRoot":"","sources":["../../src/runtime/text.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,wBAAgB,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAoB5C"}

package/dist/runtime/text.js

@@ -1,25 +0,0 @@
-/**
- * Removes the common leading indentation from a multi-line string.
- *
- * Steps are written as template literals in a spec file, so `action` and `expect` arrive
- * carrying whatever indentation the surrounding code happened to have. That indentation is
- * an artifact of the source layout, not of the step, and it should not reach the report or
- * the failure message.
- */
-export function dedent(value) {
-    const lines = value.split('\n').map((line) => line.replace(/\s+$/, ''));
-    while (lines.length > 0 && lines[0].trim() === '') {
-        lines.shift();
-    }
-    while (lines.length > 0 && lines[lines.length - 1].trim() === '') {
-        lines.pop();
-    }
-    const indent = lines
-        .filter((line) => line.trim() !== '')
-        .reduce((smallest, line) => Math.min(smallest, line.length - line.trimStart().length), Number.MAX_SAFE_INTEGER);
-    if (indent === Number.MAX_SAFE_INTEGER) {
-        return '';
-    }
-    return lines.map((line) => line.slice(indent)).join('\n');
-}
-//# sourceMappingURL=text.js.map

package/dist/runtime/text.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"text.js","sourceRoot":"","sources":["../../src/runtime/text.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,MAAM,UAAU,MAAM,CAAC,KAAa;IACnC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAC,CAAA;IAEvE,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;QACnD,KAAK,CAAC,KAAK,EAAE,CAAA;IACd,CAAC;IAED,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;QAClE,KAAK,CAAC,GAAG,EAAE,CAAA;IACZ,CAAC;IAED,MAAM,MAAM,GAAG,KAAK;SAClB,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC;SACpC,MAAM,CAAC,CAAC,QAAQ,EAAE,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,QAAQ,EAAE,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC,gBAAgB,CAAC,CAAA;IAEhH,IAAI,MAAM,KAAK,MAAM,CAAC,gBAAgB,EAAE,CAAC;QACxC,OAAO,EAAE,CAAA;IACV,CAAC;IAED,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AAC1D,CAAC"}

package/dist/runtime/types.d.ts

@@ -1,53 +0,0 @@
-import type { ChatCompletionMessageParam } from 'openai/resources/chat/completions';
-import type { Diagnostic, Usage } from '../contracts/types.js';
-import type { StepIntent } from '../driver.js';
-import type { ToolExecution } from '../tools/types.js';
-export type Step = Pick<StepIntent, 'action' | 'expect'>;
-export type StepAssertion = {
-    passed: boolean;
-    actual: string;
-};
-export type InternalTerminationReason = 'met-expectation' | 'failed-expectation' | 'loop-detected' | 'turn-cap-exceeded' | 'step-timeout' | 'scenario-timeout' | 'tool-error' | 'provider-error' | 'token-budget-exceeded' | 'interrupted' | 'internal-error';
-export type InternalStepToolCall = {
-    turn: number;
-    driverId: string;
-    name: string;
-    arguments: unknown;
-    status: 'ok' | 'error';
-};
-export type InternalStepReport = {
-    step: StepIntent;
-    outcome: 'passed' | 'failed';
-    category: StepCategory;
-    reason: InternalTerminationReason;
-    actual?: string;
-    turns: number;
-    durationMs: number;
-    usage: Usage;
-    toolCalls: InternalStepToolCall[];
-    transcript: TranscriptEntry[];
-    diagnostics: Diagnostic[];
-};
-export type StepCategory = 'app' | 'model' | 'infra';
-export type TranscriptEntry = {
-    turn: number;
-    role: 'assistant' | 'tool';
-    content: string;
-};
-export type ContextMessage = {
-    message: ChatCompletionMessageParam;
-    ephemeral?: boolean;
-};
-export type TurnOutcome = {
-    kind: 'continue';
-    toolResults: ToolExecution[];
-    messages: ChatCompletionMessageParam[];
-} | {
-    kind: 'assertion';
-    passed: boolean;
-    actual: string;
-} | {
-    kind: 'stuck';
-    reason: 'loop-detected';
-};
-//# sourceMappingURL=types.d.ts.map

package/dist/runtime/types.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/runtime/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,0BAA0B,EAAE,MAAM,mCAAmC,CAAA;AACnF,OAAO,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,uBAAuB,CAAA;AAC9D,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,cAAc,CAAA;AAC9C,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAEtD,MAAM,MAAM,IAAI,GAAG,IAAI,CAAC,UAAU,EAAE,QAAQ,GAAG,QAAQ,CAAC,CAAA;AAExD,MAAM,MAAM,aAAa,GAAG;IAAE,MAAM,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,CAAA;AAE/D,MAAM,MAAM,yBAAyB,GAClC,iBAAiB,GACjB,oBAAoB,GACpB,eAAe,GACf,mBAAmB,GACnB,cAAc,GACd,kBAAkB,GAClB,YAAY,GACZ,gBAAgB,GAChB,uBAAuB,GACvB,aAAa,GACb,gBAAgB,CAAA;AAEnB,MAAM,MAAM,oBAAoB,GAAG;IAClC,IAAI,EAAE,MAAM,CAAA;IACZ,QAAQ,EAAE,MAAM,CAAA;IAChB,IAAI,EAAE,MAAM,CAAA;IACZ,SAAS,EAAE,OAAO,CAAA;IAClB,MAAM,EAAE,IAAI,GAAG,OAAO,CAAA;CACtB,CAAA;AAED,MAAM,MAAM,kBAAkB,GAAG;IAChC,IAAI,EAAE,UAAU,CAAA;IAChB,OAAO,EAAE,QAAQ,GAAG,QAAQ,CAAA;IAC5B,QAAQ,EAAE,YAAY,CAAA;IACtB,MAAM,EAAE,yBAAyB,CAAA;IACjC,MAAM,CAAC,EAAE,MAAM,CAAA;IACf,KAAK,EAAE,MAAM,CAAA;IACb,UAAU,EAAE,MAAM,CAAA;IAClB,KAAK,EAAE,KAAK,CAAA;IACZ,SAAS,EAAE,oBAAoB,EAAE,CAAA;IACjC,UAAU,EAAE,eAAe,EAAE,CAAA;IAC7B,WAAW,EAAE,UAAU,EAAE,CAAA;CACzB,CAAA;AAED,MAAM,MAAM,YAAY,GAAG,KAAK,GAAG,OAAO,GAAG,OAAO,CAAA;AAEpD,MAAM,MAAM,eAAe,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,WAAW,GAAG,MAAM,CAAC;IAAC,OAAO,EAAE,MAAM,CAAA;CAAE,CAAA;AAE3F,MAAM,MAAM,cAAc,GAAG;IAAE,OAAO,EAAE,0BAA0B,CAAC;IAAC,SAAS,CAAC,EAAE,OAAO,CAAA;CAAE,CAAA;AAEzF,MAAM,MAAM,WAAW,GACpB;IAAE,IAAI,EAAE,UAAU,CAAC;IAAC,WAAW,EAAE,aAAa,EAAE,CAAC;IAAC,QAAQ,EAAE,0BAA0B,EAAE,CAAA;CAAE,GAC1F;IAAE,IAAI,EAAE,WAAW,CAAC;IAAC,MAAM,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,GACtD;IAAE,IAAI,EAAE,OAAO,CAAC;IAAC,MAAM,EAAE,eAAe,CAAA;CAAE,CAAA"}

package/dist/runtime/types.js

@@ -1,2 +0,0 @@
-export {};
-//# sourceMappingURL=types.js.map

package/dist/runtime/types.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/runtime/types.ts"],"names":[],"mappings":""}

package/dist/runtime/usage-tracker.d.ts

@@ -1,37 +0,0 @@
-import type { ChatCompletion } from 'openai/resources/chat/completions';
-import type { Usage } from '../contracts/types.js';
-export type UsageCheckpoint = Readonly<{
-    promptTokens: number;
-    cachedPromptTokens: number;
-    completionTokens: number;
-    reportedResponses: number;
-    missingResponses: number;
-}>;
-export declare class TokenBudgetExceededError extends Error {
-    readonly budgetTokens: number;
-    readonly totalTokens: number;
-    constructor(budgetTokens: number, totalTokens: number);
-}
-export declare class ProviderUsageUnavailableError extends Error {
-    constructor();
-}
-export declare class ProviderUsageInvalidError extends Error {
-    constructor(message: string);
-}
-export declare class ScenarioUsageTracker {
-    private readonly budgetTokens?;
-    private promptTokens;
-    private cachedPromptTokens;
-    private completionTokens;
-    private reportedResponses;
-    private missingResponses;
-    constructor(budgetTokens?: number);
-    beginStep(): UsageCheckpoint;
-    record(usage: ChatCompletion['usage']): void;
-    stepUsage(checkpoint: UsageCheckpoint): Usage;
-    usage(): Usage & {
-        state: 'complete' | 'partial' | 'unavailable';
-    };
-    private checkpoint;
-}
-//# sourceMappingURL=usage-tracker.d.ts.map

package/dist/runtime/usage-tracker.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"usage-tracker.d.ts","sourceRoot":"","sources":["../../src/runtime/usage-tracker.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,mCAAmC,CAAA;AACvE,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,uBAAuB,CAAA;AAElD,MAAM,MAAM,eAAe,GAAG,QAAQ,CAAC;IACtC,YAAY,EAAE,MAAM,CAAA;IACpB,kBAAkB,EAAE,MAAM,CAAA;IAC1B,gBAAgB,EAAE,MAAM,CAAA;IACxB,iBAAiB,EAAE,MAAM,CAAA;IACzB,gBAAgB,EAAE,MAAM,CAAA;CACxB,CAAC,CAAA;AAEF,qBAAa,wBAAyB,SAAQ,KAAK;IAEjD,QAAQ,CAAC,YAAY,EAAE,MAAM;IAC7B,QAAQ,CAAC,WAAW,EAAE,MAAM;gBADnB,YAAY,EAAE,MAAM,EACpB,WAAW,EAAE,MAAM;CAK7B;AAED,qBAAa,6BAA8B,SAAQ,KAAK;;CAKvD;AAED,qBAAa,yBAA0B,SAAQ,KAAK;gBACvC,OAAO,EAAE,MAAM;CAI3B;AAED,qBAAa,oBAAoB;IAOpB,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAC;IAN1C,OAAO,CAAC,YAAY,CAAI;IACxB,OAAO,CAAC,kBAAkB,CAAI;IAC9B,OAAO,CAAC,gBAAgB,CAAI;IAC5B,OAAO,CAAC,iBAAiB,CAAI;IAC7B,OAAO,CAAC,gBAAgB,CAAI;gBAEC,YAAY,CAAC,EAAE,MAAM;IAMlD,SAAS,IAAI,eAAe;IAI5B,MAAM,CAAC,KAAK,EAAE,cAAc,CAAC,OAAO,CAAC,GAAG,IAAI;IAgC5C,SAAS,CAAC,UAAU,EAAE,eAAe,GAAG,KAAK;IAQ7C,KAAK,IAAI,KAAK,GAAG;QAAE,KAAK,EAAE,UAAU,GAAG,SAAS,GAAG,aAAa,CAAA;KAAE;IAYlE,OAAO,CAAC,UAAU;CASlB"}

package/dist/runtime/usage-tracker.js

@@ -1,110 +0,0 @@
-export class TokenBudgetExceededError extends Error {
-    budgetTokens;
-    totalTokens;
-    constructor(budgetTokens, totalTokens) {
-        super(`Scenario token budget of ${budgetTokens} was exceeded by response total ${totalTokens}`);
-        this.budgetTokens = budgetTokens;
-        this.totalTokens = totalTokens;
-        this.name = 'TokenBudgetExceededError';
-    }
-}
-export class ProviderUsageUnavailableError extends Error {
-    constructor() {
-        super('Provider response omitted usage while a scenario token budget is active');
-        this.name = 'ProviderUsageUnavailableError';
-    }
-}
-export class ProviderUsageInvalidError extends Error {
-    constructor(message) {
-        super(`Provider response contained invalid usage: ${message}`);
-        this.name = 'ProviderUsageInvalidError';
-    }
-}
-export class ScenarioUsageTracker {
-    budgetTokens;
-    promptTokens = 0;
-    cachedPromptTokens = 0;
-    completionTokens = 0;
-    reportedResponses = 0;
-    missingResponses = 0;
-    constructor(budgetTokens) {
-        this.budgetTokens = budgetTokens;
-        if (budgetTokens !== undefined && (!Number.isSafeInteger(budgetTokens) || budgetTokens < 1)) {
-            throw new Error('Scenario token budget must be a positive integer');
-        }
-    }
-    beginStep() {
-        return this.checkpoint();
-    }
-    record(usage) {
-        if (!usage) {
-            this.missingResponses++;
-            if (this.budgetTokens !== undefined)
-                throw new ProviderUsageUnavailableError();
-            return;
-        }
-        const promptTokens = requiredTokenCount(usage.prompt_tokens, 'prompt_tokens');
-        const completionTokens = requiredTokenCount(usage.completion_tokens, 'completion_tokens');
-        const totalTokens = requiredTokenCount(usage.total_tokens, 'total_tokens');
-        if (totalTokens !== promptTokens + completionTokens) {
-            throw new ProviderUsageInvalidError('total_tokens must equal prompt_tokens plus completion_tokens');
-        }
-        const cachedPromptTokens = optionalTokenCount(usage.prompt_tokens_details?.cached_tokens, 'prompt_tokens_details.cached_tokens');
-        if (cachedPromptTokens > promptTokens) {
-            throw new ProviderUsageInvalidError('cached_tokens cannot exceed prompt_tokens');
-        }
-        this.promptTokens += promptTokens;
-        this.cachedPromptTokens += cachedPromptTokens;
-        this.completionTokens += completionTokens;
-        this.reportedResponses++;
-        const scenarioTotalTokens = this.promptTokens + this.completionTokens;
-        if (this.budgetTokens !== undefined && scenarioTotalTokens > this.budgetTokens) {
-            throw new TokenBudgetExceededError(this.budgetTokens, scenarioTotalTokens);
-        }
-    }
-    stepUsage(checkpoint) {
-        return usageFrom({
-            promptTokens: this.promptTokens - checkpoint.promptTokens,
-            cachedPromptTokens: this.cachedPromptTokens - checkpoint.cachedPromptTokens,
-            completionTokens: this.completionTokens - checkpoint.completionTokens,
-        });
-    }
-    usage() {
-        const state = this.reportedResponses === 0 ? 'unavailable' : this.missingResponses > 0 ? 'partial' : 'complete';
-        return {
-            ...usageFrom({
-                promptTokens: this.promptTokens,
-                cachedPromptTokens: this.cachedPromptTokens,
-                completionTokens: this.completionTokens,
-            }),
-            state,
-        };
-    }
-    checkpoint() {
-        return {
-            promptTokens: this.promptTokens,
-            cachedPromptTokens: this.cachedPromptTokens,
-            completionTokens: this.completionTokens,
-            reportedResponses: this.reportedResponses,
-            missingResponses: this.missingResponses,
-        };
-    }
-}
-function requiredTokenCount(value, field) {
-    if (!Number.isSafeInteger(value) || value < 0) {
-        throw new ProviderUsageInvalidError(`${field} must be a non-negative safe integer`);
-    }
-    return value;
-}
-function optionalTokenCount(value, field) {
-    return value === undefined ? 0 : requiredTokenCount(value, field);
-}
-function usageFrom(input) {
-    return {
-        promptTokens: input.promptTokens,
-        cachedPromptTokens: Math.min(input.cachedPromptTokens, input.promptTokens),
-        completionTokens: input.completionTokens,
-        totalTokens: input.promptTokens + input.completionTokens,
-    };
-}
-//# sourceMappingURL=usage-tracker.js.map

package/dist/runtime/usage-tracker.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"usage-tracker.js","sourceRoot":"","sources":["../../src/runtime/usage-tracker.ts"],"names":[],"mappings":"AAWA,MAAM,OAAO,wBAAyB,SAAQ,KAAK;IAExC;IACA;IAFV,YACU,YAAoB,EACpB,WAAmB;QAE5B,KAAK,CAAC,4BAA4B,YAAY,mCAAmC,WAAW,EAAE,CAAC,CAAA;QAHtF,iBAAY,GAAZ,YAAY,CAAQ;QACpB,gBAAW,GAAX,WAAW,CAAQ;QAG5B,IAAI,CAAC,IAAI,GAAG,0BAA0B,CAAA;IACvC,CAAC;CACD;AAED,MAAM,OAAO,6BAA8B,SAAQ,KAAK;IACvD;QACC,KAAK,CAAC,yEAAyE,CAAC,CAAA;QAChF,IAAI,CAAC,IAAI,GAAG,+BAA+B,CAAA;IAC5C,CAAC;CACD;AAED,MAAM,OAAO,yBAA0B,SAAQ,KAAK;IACnD,YAAY,OAAe;QAC1B,KAAK,CAAC,8CAA8C,OAAO,EAAE,CAAC,CAAA;QAC9D,IAAI,CAAC,IAAI,GAAG,2BAA2B,CAAA;IACxC,CAAC;CACD;AAED,MAAM,OAAO,oBAAoB;IAOH;IANrB,YAAY,GAAG,CAAC,CAAA;IAChB,kBAAkB,GAAG,CAAC,CAAA;IACtB,gBAAgB,GAAG,CAAC,CAAA;IACpB,iBAAiB,GAAG,CAAC,CAAA;IACrB,gBAAgB,GAAG,CAAC,CAAA;IAE5B,YAA6B,YAAqB;QAArB,iBAAY,GAAZ,YAAY,CAAS;QACjD,IAAI,YAAY,KAAK,SAAS,IAAI,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,YAAY,CAAC,IAAI,YAAY,GAAG,CAAC,CAAC,EAAE,CAAC;YAC7F,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;QACpE,CAAC;IACF,CAAC;IAED,SAAS;QACR,OAAO,IAAI,CAAC,UAAU,EAAE,CAAA;IACzB,CAAC;IAED,MAAM,CAAC,KAA8B;QACpC,IAAI,CAAC,KAAK,EAAE,CAAC;YACZ,IAAI,CAAC,gBAAgB,EAAE,CAAA;YACvB,IAAI,IAAI,CAAC,YAAY,KAAK,SAAS;gBAAE,MAAM,IAAI,6BAA6B,EAAE,CAAA;YAC9E,OAAM;QACP,CAAC;QAED,MAAM,YAAY,GAAG,kBAAkB,CAAC,KAAK,CAAC,aAAa,EAAE,eAAe,CAAC,CAAA;QAC7E,MAAM,gBAAgB,GAAG,kBAAkB,CAAC,KAAK,CAAC,iBAAiB,EAAE,mBAAmB,CAAC,CAAA;QACzF,MAAM,WAAW,GAAG,kBAAkB,CAAC,KAAK,CAAC,YAAY,EAAE,cAAc,CAAC,CAAA;QAC1E,IAAI,WAAW,KAAK,YAAY,GAAG,gBAAgB,EAAE,CAAC;YACrD,MAAM,IAAI,yBAAyB,CAAC,8DAA8D,CAAC,CAAA;QACpG,CAAC;QAED,MAAM,kBAAkB,GAAG,kBAAkB,CAC5C,KAAK,CAAC,qBAAqB,EAAE,aAAa,EAC1C,qCAAqC,CACrC,CAAA;QACD,IAAI,kBAAkB,GAAG,YAAY,EAAE,CAAC;YACvC,MAAM,IAAI,yBAAyB,CAAC,2CAA2C,CAAC,CAAA;QACjF,CAAC;QACD,IAAI,CAAC,YAAY,IAAI,YAAY,CAAA;QACjC,IAAI,CAAC,kBAAkB,IAAI,kBAAkB,CAAA;QAC7C,IAAI,CAAC,gBAAgB,IAAI,gBAAgB,CAAA;QACzC,IAAI,CAAC,iBAAiB,EAAE,CAAA;QAExB,MAAM,mBAAmB,GAAG,IAAI,CAAC,YAAY,GAAG,IAAI,CAAC,gBAAgB,CAAA;QACrE,IAAI,IAAI,CAAC,YAAY,KAAK,SAAS,IAAI,mBAAmB,GAAG,IAAI,CAAC,YAAY,EAAE,CAAC;YAChF,MAAM,IAAI,wBAAwB,CAAC,IAAI,CAAC,YAAY,EAAE,mBAAmB,CAAC,CAAA;QAC3E,CAAC;IACF,CAAC;IAED,SAAS,CAAC,UAA2B;QACpC,OAAO,SAAS,CAAC;YAChB,YAAY,EAAE,IAAI,CAAC,YAAY,GAAG,UAAU,CAAC,YAAY;YACzD,kBAAkB,EAAE,IAAI,CAAC,kBAAkB,GAAG,UAAU,CAAC,kBAAkB;YAC3E,gBAAgB,EAAE,IAAI,CAAC,gBAAgB,GAAG,UAAU,CAAC,gBAAgB;SACrE,CAAC,CAAA;IACH,CAAC;IAED,KAAK;QACJ,MAAM,KAAK,GAAG,IAAI,CAAC,iBAAiB,KAAK,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,IAAI,CAAC,gBAAgB,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,UAAU,CAAA;QAC/G,OAAO;YACN,GAAG,SAAS,CAAC;gBACZ,YAAY,EAAE,IAAI,CAAC,YAAY;gBAC/B,kBAAkB,EAAE,IAAI,CAAC,kBAAkB;gBAC3C,gBAAgB,EAAE,IAAI,CAAC,gBAAgB;aACvC,CAAC;YACF,KAAK;SACL,CAAA;IACF,CAAC;IAEO,UAAU;QACjB,OAAO;YACN,YAAY,EAAE,IAAI,CAAC,YAAY;YAC/B,kBAAkB,EAAE,IAAI,CAAC,kBAAkB;YAC3C,gBAAgB,EAAE,IAAI,CAAC,gBAAgB;YACvC,iBAAiB,EAAE,IAAI,CAAC,iBAAiB;YACzC,gBAAgB,EAAE,IAAI,CAAC,gBAAgB;SACvC,CAAA;IACF,CAAC;CACD;AAED,SAAS,kBAAkB,CAAC,KAAc,EAAE,KAAa;IACxD,IAAI,CAAC,MAAM,CAAC,aAAa,CAAC,KAAK,CAAC,IAAK,KAAgB,GAAG,CAAC,EAAE,CAAC;QAC3D,MAAM,IAAI,yBAAyB,CAAC,GAAG,KAAK,sCAAsC,CAAC,CAAA;IACpF,CAAC;IACD,OAAO,KAAe,CAAA;AACvB,CAAC;AAED,SAAS,kBAAkB,CAAC,KAAc,EAAE,KAAa;IACxD,OAAO,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,kBAAkB,CAAC,KAAK,EAAE,KAAK,CAAC,CAAA;AAClE,CAAC;AAED,SAAS,SAAS,CAAC,KAAqF;IACvG,OAAO;QACN,YAAY,EAAE,KAAK,CAAC,YAAY;QAChC,kBAAkB,EAAE,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,kBAAkB,EAAE,KAAK,CAAC,YAAY,CAAC;QAC1E,gBAAgB,EAAE,KAAK,CAAC,gBAAgB;QACxC,WAAW,EAAE,KAAK,CAAC,YAAY,GAAG,KAAK,CAAC,gBAAgB;KACxD,CAAA;AACF,CAAC"}

package/dist/tools/define-agent-tool.d.ts

@@ -1,20 +0,0 @@
-import { z } from 'zod/v4';
-import { AgentTool, AgentToolContext, AgentToolResult } from './types.js';
-type ToolConfig<TSchema extends z.ZodType> = {
-    name: string;
-    description: string;
-    schema: TSchema;
-    strict?: boolean;
-    handler: (args: z.infer<TSchema>, context: AgentToolContext) => Promise<AgentToolResult> | AgentToolResult;
-};
-/**
- * Creates an internal runner tool, which may return a step assertion.
- * Driver authors use defineDriverTool from '@xoxoai/checkmate/driver' instead.
- */
-export declare function defineAgentTool<TSchema extends z.ZodType>(toolConfig: ToolConfig<TSchema>): AgentTool;
-/**
- * Internal shorthand for defineAgentTool; not part of the public driver API.
- */
-export declare const defineTool: typeof defineAgentTool;
-export {};
-//# sourceMappingURL=define-agent-tool.d.ts.map

package/dist/tools/define-agent-tool.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"define-agent-tool.d.ts","sourceRoot":"","sources":["../../src/tools/define-agent-tool.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,QAAQ,CAAA;AAC1B,OAAO,EAAE,SAAS,EAAE,gBAAgB,EAAuB,eAAe,EAAE,MAAM,YAAY,CAAA;AAE9F,KAAK,UAAU,CAAC,OAAO,SAAS,CAAC,CAAC,OAAO,IAAI;IAC5C,IAAI,EAAE,MAAM,CAAA;IACZ,WAAW,EAAE,MAAM,CAAA;IACnB,MAAM,EAAE,OAAO,CAAA;IACf,MAAM,CAAC,EAAE,OAAO,CAAA;IAChB,OAAO,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,OAAO,EAAE,gBAAgB,KAAK,OAAO,CAAC,eAAe,CAAC,GAAG,eAAe,CAAA;CAC1G,CAAA;AAED;;;GAGG;AACH,wBAAgB,eAAe,CAAC,OAAO,SAAS,CAAC,CAAC,OAAO,EAAE,UAAU,EAAE,UAAU,CAAC,OAAO,CAAC,GAAG,SAAS,CAsBrG;AAED;;GAEG;AACH,eAAO,MAAM,UAAU,wBAAkB,CAAA"}