@swarmdo/cli 1.4.8

package/dist/src/benchmarks/gaia-critic.d.ts

@@ -0,0 +1,123 @@
+/**
+ * GAIA Adversarial Critic — ADR-135 Track D
+ *
+ * After the main agent produces a candidate answer, a Sonnet pass reviews it.
+ * If the critic verdict is "fail", the orchestrator re-runs the agent with the
+ * critique injected as additional context.
+ *
+ * Motivation (iter 29 finding): tool quality is the bottleneck on L1 (~20.8%).
+ * The critic catches wrong-because-of-bad-tool-result answers BEFORE submission,
+ * without requiring better search backends.  Expected L1 lift: +3-5pp.
+ *
+ * Design:
+ *  - NEW file only; NOT wired into gaia-bench.ts yet to avoid merge conflicts
+ *    with in-flight iter 29/31/34 branches.  Wiring is a small follow-up PR.
+ *  - `criticReview()` — single Sonnet call, returns structured verdict.
+ *  - `runGaiaAgentWithCritic()` — orchestration wrapper: runs agent, calls
+ *    critic, retries once on "fail".  "uncertain" is treated as "pass" (don't
+ *    burn retries on borderline cases).
+ *  - API errors and malformed JSON from the critic are caught; original
+ *    candidate is returned with an error-flagged verdict.
+ *  - Default opt-in: enableCritic=false.  Set true via RunWithCriticOptions.
+ *
+ * Cost discipline:
+ *  - Critic uses claude-sonnet-4-6 (separate from the agent's default Haiku).
+ *  - One critic call + one optional retry = max 2 extra Sonnet calls per Q.
+ *  - Approximate extra cost per question: ~$0.003-0.005 (well within budget).
+ *
+ * Plugin sync TODO (follow-up PR after gaia-bench wiring):
+ *  - Update plugins/swarmdo-workflows/commands/gaia-run.md with --enable-critic flag.
+ *  - Update plugins/swarmdo-workflows/skills/gaia-debugging/SKILL.md: add critic
+ *    as a recommended diagnostic step for wrong-answer analysis.
+ *
+ * Refs: ADR-135, ADR-133, iter 29 finding, #2156
+ */
+import { GaiaQuestion } from './gaia-loader.js';
+import { GaiaAgentResult, GaiaAgentOptions } from './gaia-agent.js';
+/**
+ * Structured verdict returned by the adversarial critic.
+ *
+ * - "pass"      → critic agrees with the candidate; no retry needed.
+ * - "fail"      → critic found a flaw; suggestedRevision provided if possible.
+ * - "uncertain" → critic is unsure; treated as "pass" to avoid burning retries.
+ */
+export type CriticVerdictType = 'pass' | 'fail' | 'uncertain';
+export interface CriticVerdict {
+    /** Critic's assessment of the candidate answer. */
+    verdict: CriticVerdictType;
+    /** Short explanation of the reasoning (1-3 sentences). */
+    reasoning: string;
+    /** Suggested correction when verdict is "fail"; may be empty string. */
+    suggestedRevision?: string;
+    /** Estimated USD cost of this critic call. */
+    costUsd: number;
+    /** True when the critic call itself failed (API error, malformed JSON). */
+    error?: boolean;
+    /** Original raw response text when parse failed. */
+    rawResponse?: string;
+}
+export interface CriticOptions {
+    /**
+     * Model to use for the critic pass.
+     * Defaults to 'claude-sonnet-4-6' (intentionally separate from agent model).
+     */
+    model?: string;
+    /**
+     * Maximum number of agent retries on "fail" verdict.
+     * Default: 1 (one retry).  Set to 0 to disable retries (observe-only mode).
+     */
+    maxRetries?: number;
+    /** Anthropic API key (resolved via env/gcloud if omitted). */
+    apiKey?: string;
+}
+export interface GaiaAgentResultWithCritic extends GaiaAgentResult {
+    /** All critic verdicts collected (one per agent attempt). */
+    criticVerdicts: CriticVerdict[];
+    /** How many retries were actually attempted (0 if critic passed first time). */
+    retriesAttempted: number;
+}
+/** Options for runGaiaAgentWithCritic, extending core agent options. */
+export interface RunWithCriticOptions extends GaiaAgentOptions {
+    /**
+     * Enable the adversarial critic pass.  Default: false.
+     * When false, runGaiaAgentWithCritic behaves identically to runGaiaAgent.
+     */
+    enableCritic?: boolean;
+    /** Critic-specific configuration. */
+    criticOptions?: CriticOptions;
+}
+/**
+ * Run the adversarial critic against a candidate answer.
+ *
+ * @param question     - The GAIA question being evaluated.
+ * @param candidateAnswer - The agent's proposed final answer.
+ * @param trajectory   - Lightweight trajectory summary from the agent run.
+ * @param options      - Critic configuration (model, apiKey).
+ * @returns            CriticVerdict with verdict, reasoning, cost.
+ */
+export declare function criticReview(question: GaiaQuestion, candidateAnswer: string, trajectory: {
+    steps: Array<{
+        tool?: string;
+        result?: string;
+        reasoning?: string;
+    }>;
+    turns: number;
+}, options?: CriticOptions): Promise<CriticVerdict>;
+/**
+ * Run the GAIA agent with an optional adversarial critic pass.
+ *
+ * When enableCritic=false (default), this is a thin pass-through to
+ * runGaiaAgent with an empty criticVerdicts array.
+ *
+ * When enableCritic=true:
+ *  1. Run runGaiaAgent normally.
+ *  2. If the agent produced a finalAnswer, call criticReview.
+ *  3. If verdict is "fail" and retriesAttempted < maxRetries:
+ *     a. Re-run the agent with the critique injected as additional context.
+ *     b. Call criticReview again on the new answer.
+ *  4. Return the final result with all critic verdicts attached.
+ *
+ * Note on "uncertain": treated as "pass" (no retry triggered).
+ */
+export declare function runGaiaAgentWithCritic(question: GaiaQuestion, options?: RunWithCriticOptions): Promise<GaiaAgentResultWithCritic>;
+//# sourceMappingURL=gaia-critic.d.ts.map

package/dist/src/benchmarks/gaia-critic.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"gaia-critic.d.ts","sourceRoot":"","sources":["../../../src/benchmarks/gaia-critic.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AAGH,OAAO,EACL,YAAY,EACb,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EAEL,eAAe,EACf,gBAAgB,EACjB,MAAM,iBAAiB,CAAC;AAwBzB;;;;;;GAMG;AACH,MAAM,MAAM,iBAAiB,GAAG,MAAM,GAAG,MAAM,GAAG,WAAW,CAAC;AAE9D,MAAM,WAAW,aAAa;IAC5B,mDAAmD;IACnD,OAAO,EAAE,iBAAiB,CAAC;IAC3B,0DAA0D;IAC1D,SAAS,EAAE,MAAM,CAAC;IAClB,wEAAwE;IACxE,iBAAiB,CAAC,EAAE,MAAM,CAAC;IAC3B,8CAA8C;IAC9C,OAAO,EAAE,MAAM,CAAC;IAChB,2EAA2E;IAC3E,KAAK,CAAC,EAAE,OAAO,CAAC;IAChB,oDAAoD;IACpD,WAAW,CAAC,EAAE,MAAM,CAAC;CACtB;AAED,MAAM,WAAW,aAAa;IAC5B;;;OAGG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC;IACf;;;OAGG;IACH,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,8DAA8D;IAC9D,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAGD,MAAM,WAAW,yBAA0B,SAAQ,eAAe;IAChE,6DAA6D;IAC7D,cAAc,EAAE,aAAa,EAAE,CAAC;IAChC,gFAAgF;IAChF,gBAAgB,EAAE,MAAM,CAAC;CAC1B;AAED,wEAAwE;AACxE,MAAM,WAAW,oBAAqB,SAAQ,gBAAgB;IAC5D;;;OAGG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB,qCAAqC;IACrC,aAAa,CAAC,EAAE,aAAa,CAAC;CAC/B;AAqID;;;;;;;;GAQG;AACH,wBAAsB,YAAY,CAChC,QAAQ,EAAE,YAAY,EACtB,eAAe,EAAE,MAAM,EACvB,UAAU,EAAE;IAAE,KAAK,EAAE,KAAK,CAAC;QAAE,IAAI,CAAC,EAAE,MAAM,CAAC;QAAC,MAAM,CAAC,EAAE,MAAM,CAAC;QAAC,SAAS,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC,CAAC;IAAC,KAAK,EAAE,MAAM,CAAA;CAAE,EACnG,OAAO,CAAC,EAAE,aAAa,GACtB,OAAO,CAAC,aAAa,CAAC,CAyExB;AAMD;;;;;;;;;;;;;;;GAeG;AACH,wBAAsB,sBAAsB,CAC1C,QAAQ,EAAE,YAAY,EACtB,OAAO,GAAE,oBAAyB,GACjC,OAAO,CAAC,yBAAyB,CAAC,CA4EpC"}

package/dist/src/benchmarks/gaia-critic.js

@@ -0,0 +1,312 @@
+/**
+ * GAIA Adversarial Critic — ADR-135 Track D
+ *
+ * After the main agent produces a candidate answer, a Sonnet pass reviews it.
+ * If the critic verdict is "fail", the orchestrator re-runs the agent with the
+ * critique injected as additional context.
+ *
+ * Motivation (iter 29 finding): tool quality is the bottleneck on L1 (~20.8%).
+ * The critic catches wrong-because-of-bad-tool-result answers BEFORE submission,
+ * without requiring better search backends.  Expected L1 lift: +3-5pp.
+ *
+ * Design:
+ *  - NEW file only; NOT wired into gaia-bench.ts yet to avoid merge conflicts
+ *    with in-flight iter 29/31/34 branches.  Wiring is a small follow-up PR.
+ *  - `criticReview()` — single Sonnet call, returns structured verdict.
+ *  - `runGaiaAgentWithCritic()` — orchestration wrapper: runs agent, calls
+ *    critic, retries once on "fail".  "uncertain" is treated as "pass" (don't
+ *    burn retries on borderline cases).
+ *  - API errors and malformed JSON from the critic are caught; original
+ *    candidate is returned with an error-flagged verdict.
+ *  - Default opt-in: enableCritic=false.  Set true via RunWithCriticOptions.
+ *
+ * Cost discipline:
+ *  - Critic uses claude-sonnet-4-6 (separate from the agent's default Haiku).
+ *  - One critic call + one optional retry = max 2 extra Sonnet calls per Q.
+ *  - Approximate extra cost per question: ~$0.003-0.005 (well within budget).
+ *
+ * Plugin sync TODO (follow-up PR after gaia-bench wiring):
+ *  - Update plugins/swarmdo-workflows/commands/gaia-run.md with --enable-critic flag.
+ *  - Update plugins/swarmdo-workflows/skills/gaia-debugging/SKILL.md: add critic
+ *    as a recommended diagnostic step for wrong-answer analysis.
+ *
+ * Refs: ADR-135, ADR-133, iter 29 finding, #2156
+ */
+import { execSync } from 'node:child_process';
+import { runGaiaAgent, } from './gaia-agent.js';
+// ---------------------------------------------------------------------------
+// Constants
+// ---------------------------------------------------------------------------
+const ANTHROPIC_API_URL = 'https://api.anthropic.com/v1/messages';
+const ANTHROPIC_API_VERSION = '2023-06-01';
+/** Default model for the critic — Sonnet for higher reasoning quality. */
+const DEFAULT_CRITIC_MODEL = 'claude-sonnet-4-6';
+/** Max tokens for the critic response (verdict JSON is short). */
+const CRITIC_MAX_TOKENS = 512;
+/** Sonnet pricing (input/output per million tokens, 2026-05-27).
+ *  Used only for cost estimation in CriticVerdict. */
+const SONNET_INPUT_COST_PER_M = 3.0;
+const SONNET_OUTPUT_COST_PER_M = 15.0;
+// ---------------------------------------------------------------------------
+// Internal helpers
+// ---------------------------------------------------------------------------
+/** Resolve Anthropic API key using the same precedence as gaia-agent.ts. */
+function resolveApiKey(override) {
+    if (override)
+        return override;
+    const fromEnv = process.env['ANTHROPIC_API_KEY'];
+    if (fromEnv)
+        return fromEnv;
+    try {
+        return execSync('gcloud secrets versions access latest --secret=ANTHROPIC_API_KEY', { encoding: 'utf8', stdio: ['pipe', 'pipe', 'pipe'] }).trim();
+    }
+    catch {
+        throw new Error('ANTHROPIC_API_KEY not set and gcloud fallback failed. ' +
+            'Set ANTHROPIC_API_KEY env var or ensure gcloud access.');
+    }
+}
+/**
+ * Build the critic system prompt.
+ * Instructs Sonnet to act as an adversarial reviewer and respond in JSON.
+ */
+function buildCriticSystemPrompt() {
+    return `You are an adversarial reviewer of agent answers on the GAIA benchmark. \
+Your job is to find flaws in candidate answers before they are submitted.
+
+Given a question, a candidate answer, and a summary of the agent's reasoning \
+trajectory, you must decide whether the candidate answer is correct.
+
+Evaluation criteria:
+1. Does the answer directly address what the question asks?
+2. Is there evidence in the trajectory that supports the answer?
+3. Are there obvious flaws (wrong unit, wrong format, missed constraint, \
+fabricated source, off-by-one error, truncation)?
+4. For numeric answers: is the magnitude and unit plausible?
+5. For list answers: are all required items present and correctly ordered?
+
+Respond ONLY with a JSON object — no markdown, no prose outside the JSON:
+{"verdict":"pass"|"fail"|"uncertain","reasoning":"<1-3 sentences>","suggestedRevision":"<corrected answer or empty string>"}
+
+Use "uncertain" only when you genuinely cannot determine correctness from the \
+available information. Never use "uncertain" to avoid making a decision when \
+evidence is available.`;
+}
+/**
+ * Build the critic user message combining the question, candidate answer,
+ * and a compressed view of the agent trajectory.
+ */
+function buildCriticUserMessage(question, candidateAnswer, trajectory) {
+    // Summarise trajectory: first 2 + last 2 steps to keep tokens bounded.
+    const steps = trajectory.steps ?? [];
+    const summarised = steps.length <= 4
+        ? steps
+        : [...steps.slice(0, 2), ...steps.slice(-2)];
+    const trajectoryText = summarised.length === 0
+        ? '(no tool calls recorded)'
+        : summarised.map((s, i) => {
+            const label = s.tool ? `Step ${i + 1} [${s.tool}]` : `Step ${i + 1}`;
+            const result = s.result ? ` → ${s.result.slice(0, 200)}` : '';
+            return `${label}${result}`;
+        }).join('\n');
+    return `QUESTION: ${question.question}
+
+CANDIDATE ANSWER: ${candidateAnswer}
+
+AGENT TRAJECTORY (${trajectory.turns} turns, summarised):
+${trajectoryText}`;
+}
+/**
+ * Attempt to extract a verdict from a potentially malformed JSON string.
+ * Falls back gracefully to "uncertain" with the raw text preserved.
+ */
+function parseVerdictFallback(raw) {
+    // Try standard JSON parse first.
+    try {
+        const parsed = JSON.parse(raw);
+        const verdict = parsed['verdict'];
+        if (verdict === 'pass' || verdict === 'fail' || verdict === 'uncertain') {
+            return {
+                verdict,
+                reasoning: parsed['reasoning'] ?? '',
+                suggestedRevision: parsed['suggestedRevision'] ?? '',
+            };
+        }
+    }
+    catch {
+        // Fall through to regex extraction.
+    }
+    // Regex extraction for embedded JSON in prose.
+    const jsonMatch = raw.match(/\{[\s\S]*?"verdict"\s*:\s*"(pass|fail|uncertain)"[\s\S]*?\}/);
+    if (jsonMatch) {
+        try {
+            const parsed = JSON.parse(jsonMatch[0]);
+            return {
+                verdict: parsed['verdict'],
+                reasoning: parsed['reasoning'] ?? '',
+                suggestedRevision: parsed['suggestedRevision'] ?? '',
+            };
+        }
+        catch {
+            // Fall through.
+        }
+    }
+    // Last resort: extract verdict keyword from raw text.
+    const verdictMatch = raw.match(/\b(pass|fail|uncertain)\b/i);
+    return {
+        verdict: verdictMatch
+            ? verdictMatch[1].toLowerCase()
+            : 'uncertain',
+        reasoning: `Critic returned malformed JSON; extracted verdict heuristically.`,
+        suggestedRevision: '',
+        rawResponse: raw.slice(0, 500),
+    };
+}
+// ---------------------------------------------------------------------------
+// Core critic function
+// ---------------------------------------------------------------------------
+/**
+ * Run the adversarial critic against a candidate answer.
+ *
+ * @param question     - The GAIA question being evaluated.
+ * @param candidateAnswer - The agent's proposed final answer.
+ * @param trajectory   - Lightweight trajectory summary from the agent run.
+ * @param options      - Critic configuration (model, apiKey).
+ * @returns            CriticVerdict with verdict, reasoning, cost.
+ */
+export async function criticReview(question, candidateAnswer, trajectory, options) {
+    const model = options?.model ?? DEFAULT_CRITIC_MODEL;
+    const apiKey = resolveApiKey(options?.apiKey);
+    const t0 = Date.now();
+    const requestBody = {
+        model,
+        max_tokens: CRITIC_MAX_TOKENS,
+        system: buildCriticSystemPrompt(),
+        messages: [
+            {
+                role: 'user',
+                content: buildCriticUserMessage(question, candidateAnswer, trajectory),
+            },
+        ],
+    };
+    let rawResponseText = '';
+    try {
+        const response = await fetch(ANTHROPIC_API_URL, {
+            method: 'POST',
+            headers: {
+                'Content-Type': 'application/json',
+                'x-api-key': apiKey,
+                'anthropic-version': ANTHROPIC_API_VERSION,
+            },
+            body: JSON.stringify(requestBody),
+        });
+        if (!response.ok) {
+            const errText = await response.text().catch(() => '');
+            throw new Error(`Anthropic API error ${response.status}: ${errText.slice(0, 200)}`);
+        }
+        const data = await response.json();
+        const inputTokens = data.usage?.input_tokens ?? 0;
+        const outputTokens = data.usage?.output_tokens ?? 0;
+        const costUsd = (inputTokens / 1_000_000) * SONNET_INPUT_COST_PER_M +
+            (outputTokens / 1_000_000) * SONNET_OUTPUT_COST_PER_M;
+        rawResponseText = data.content
+            .filter(b => b.type === 'text')
+            .map(b => b.text ?? '')
+            .join('');
+        const parsed = parseVerdictFallback(rawResponseText);
+        return {
+            verdict: parsed.verdict ?? 'uncertain',
+            reasoning: parsed.reasoning ?? '',
+            suggestedRevision: parsed.suggestedRevision ?? '',
+            costUsd,
+            ...(parsed.rawResponse ? { rawResponse: parsed.rawResponse } : {}),
+        };
+    }
+    catch (err) {
+        const wallMs = Date.now() - t0;
+        const errMsg = err instanceof Error ? err.message : String(err);
+        // Graceful fallback: treat critic error as "uncertain" so agent result
+        // is still returned rather than throwing.
+        return {
+            verdict: 'uncertain',
+            reasoning: `Critic call failed after ${wallMs}ms: ${errMsg.slice(0, 200)}`,
+            suggestedRevision: '',
+            costUsd: 0,
+            error: true,
+            rawResponse: rawResponseText.slice(0, 200) || undefined,
+        };
+    }
+}
+// ---------------------------------------------------------------------------
+// Orchestration wrapper
+// ---------------------------------------------------------------------------
+/**
+ * Run the GAIA agent with an optional adversarial critic pass.
+ *
+ * When enableCritic=false (default), this is a thin pass-through to
+ * runGaiaAgent with an empty criticVerdicts array.
+ *
+ * When enableCritic=true:
+ *  1. Run runGaiaAgent normally.
+ *  2. If the agent produced a finalAnswer, call criticReview.
+ *  3. If verdict is "fail" and retriesAttempted < maxRetries:
+ *     a. Re-run the agent with the critique injected as additional context.
+ *     b. Call criticReview again on the new answer.
+ *  4. Return the final result with all critic verdicts attached.
+ *
+ * Note on "uncertain": treated as "pass" (no retry triggered).
+ */
+export async function runGaiaAgentWithCritic(question, options = {}) {
+    const { enableCritic = false, criticOptions, ...agentOptions } = options;
+    // Fast path: critic disabled.
+    if (!enableCritic) {
+        const result = await runGaiaAgent(question, agentOptions);
+        return { ...result, criticVerdicts: [], retriesAttempted: 0 };
+    }
+    const maxRetries = criticOptions?.maxRetries ?? 1;
+    const criticVerdicts = [];
+    let retriesAttempted = 0;
+    // First agent run.
+    let agentResult = await runGaiaAgent(question, agentOptions);
+    // If agent timed out or errored with no answer, skip critic.
+    if (agentResult.finalAnswer == null) {
+        return { ...agentResult, criticVerdicts, retriesAttempted };
+    }
+    // Build a lightweight trajectory summary from the agent result.
+    // gaia-agent.ts doesn't expose step-level detail in GaiaAgentResult, so we
+    // synthesise from the available fields (tool call counts, turn count).
+    const makeTrajectory = (result) => ({
+        turns: result.turns,
+        steps: Object.entries(result.toolCallsByName).map(([tool, count]) => ({
+            tool,
+            result: `called ${count} time(s)`,
+        })),
+    });
+    // First critic pass.
+    let verdict = await criticReview(question, agentResult.finalAnswer, makeTrajectory(agentResult), criticOptions);
+    criticVerdicts.push(verdict);
+    // Retry loop on "fail".
+    while (verdict.verdict === 'fail' && retriesAttempted < maxRetries) {
+        retriesAttempted += 1;
+        // Inject critique as additional context via the system prompt append.
+        // We pass it through GaiaAgentOptions.criticFeedback — gaia-agent.ts does
+        // NOT yet read this field (wiring is follow-up PR), but storing it here
+        // makes the contract explicit and harmless (unknown options are ignored).
+        const retryOptions = {
+            ...agentOptions,
+            criticFeedback: `Previous answer "${agentResult.finalAnswer}" was flagged by adversarial review: ` +
+                `${verdict.reasoning}` +
+                (verdict.suggestedRevision
+                    ? ` Suggested revision: ${verdict.suggestedRevision}`
+                    : ''),
+        };
+        agentResult = await runGaiaAgent(question, retryOptions);
+        if (agentResult.finalAnswer == null) {
+            // Retry yielded no answer; stop retrying.
+            break;
+        }
+        verdict = await criticReview(question, agentResult.finalAnswer, makeTrajectory(agentResult), criticOptions);
+        criticVerdicts.push(verdict);
+    }
+    return { ...agentResult, criticVerdicts, retriesAttempted };
+}
+//# sourceMappingURL=gaia-critic.js.map

package/dist/src/benchmarks/gaia-critic.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"gaia-critic.js","sourceRoot":"","sources":["../../../src/benchmarks/gaia-critic.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiCG;AAEH,OAAO,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AAI9C,OAAO,EACL,YAAY,GAGb,MAAM,iBAAiB,CAAC;AAEzB,8EAA8E;AAC9E,YAAY;AACZ,8EAA8E;AAE9E,MAAM,iBAAiB,GAAG,uCAAuC,CAAC;AAClE,MAAM,qBAAqB,GAAG,YAAY,CAAC;AAE3C,0EAA0E;AAC1E,MAAM,oBAAoB,GAAG,mBAAmB,CAAC;AAEjD,kEAAkE;AAClE,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAE9B;sDACsD;AACtD,MAAM,uBAAuB,GAAG,GAAG,CAAC;AACpC,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAgEtC,8EAA8E;AAC9E,mBAAmB;AACnB,8EAA8E;AAE9E,4EAA4E;AAC5E,SAAS,aAAa,CAAC,QAAiB;IACtC,IAAI,QAAQ;QAAE,OAAO,QAAQ,CAAC;IAC9B,MAAM,OAAO,GAAG,OAAO,CAAC,GAAG,CAAC,mBAAmB,CAAC,CAAC;IACjD,IAAI,OAAO;QAAE,OAAO,OAAO,CAAC;IAC5B,IAAI,CAAC;QACH,OAAO,QAAQ,CACb,kEAAkE,EAClE,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CACtD,CAAC,IAAI,EAAE,CAAC;IACX,CAAC;IAAC,MAAM,CAAC;QACP,MAAM,IAAI,KAAK,CACb,wDAAwD;YACxD,wDAAwD,CACzD,CAAC;IACJ,CAAC;AACH,CAAC;AAED;;;GAGG;AACH,SAAS,uBAAuB;IAC9B,OAAO;;;;;;;;;;;;;;;;;;;uBAmBc,CAAC;AACxB,CAAC;AAED;;;GAGG;AACH,SAAS,sBAAsB,CAC7B,QAAsB,EACtB,eAAuB,EACvB,UAAmG;IAEnG,uEAAuE;IACvE,MAAM,KAAK,GAAG,UAAU,CAAC,KAAK,IAAI,EAAE,CAAC;IACrC,MAAM,UAAU,GAAG,KAAK,CAAC,MAAM,IAAI,CAAC;QAClC,CAAC,CAAC,KAAK;QACP,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAE/C,MAAM,cAAc,GAAG,UAAU,CAAC,MAAM,KAAK,CAAC;QAC5C,CAAC,CAAC,0BAA0B;QAC5B,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YACtB,MAAM,KAAK,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,IAAI,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;YACrE,MAAM,MAAM,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAC9D,OAAO,GAAG,KAAK,GAAG,MAAM,EAAE,CAAC;QAC7B,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAElB,OAAO,aAAa,QAAQ,CAAC,QAAQ;;oBAEnB,eAAe;;oBAEf,UAAU,CAAC,KAAK;EAClC,cAAc,EAAE,CAAC;AACnB,CAAC;AAED;;;GAGG;AACH,SAAS,oBAAoB,CAAC,GAAW;IACvC,iCAAiC;IACjC,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAA4B,CAAC;QAC1D,MAAM,OAAO,GAAG,MAAM,CAAC,SAAS,CAAW,CAAC;QAC5C,IAAI,OAAO,KAAK,MAAM,IAAI,OAAO,KAAK,MAAM,IAAI,OAAO,KAAK,WAAW,EAAE,CAAC;YACxE,OAAO;gBACL,OAAO;gBACP,SAAS,EAAG,MAAM,CAAC,WAAW,CAAY,IAAI,EAAE;gBAChD,iBAAiB,EAAG,MAAM,CAAC,mBAAmB,CAAY,IAAI,EAAE;aACjE,CAAC;QACJ,CAAC;IACH,CAAC;IAAC,MAAM,CAAC;QACP,oCAAoC;IACtC,CAAC;IAED,+CAA+C;IAC/C,MAAM,SAAS,GAAG,GAAG,CAAC,KAAK,CAAC,6DAA6D,CAAC,CAAC;IAC3F,IAAI,SAAS,EAAE,CAAC;QACd,IAAI,CAAC;YACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,CAA4B,CAAC;YACnE,OAAO;gBACL,OAAO,EAAE,MAAM,CAAC,SAAS,CAAsB;gBAC/C,SAAS,EAAG,MAAM,CAAC,WAAW,CAAY,IAAI,EAAE;gBAChD,iBAAiB,EAAG,MAAM,CAAC,mBAAmB,CAAY,IAAI,EAAE;aACjE,CAAC;QACJ,CAAC;QAAC,MAAM,CAAC;YACP,gBAAgB;QAClB,CAAC;IACH,CAAC;IAED,sDAAsD;IACtD,MAAM,YAAY,GAAG,GAAG,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;IAC7D,OAAO;QACL,OAAO,EAAE,YAAY;YACnB,CAAC,CAAE,YAAY,CAAC,CAAC,CAAC,CAAC,WAAW,EAAwB;YACtD,CAAC,CAAC,WAAW;QACf,SAAS,EAAE,kEAAkE;QAC7E,iBAAiB,EAAE,EAAE;QACrB,WAAW,EAAE,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC;KAC/B,CAAC;AACJ,CAAC;AAED,8EAA8E;AAC9E,uBAAuB;AACvB,8EAA8E;AAE9E;;;;;;;;GAQG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAChC,QAAsB,EACtB,eAAuB,EACvB,UAAmG,EACnG,OAAuB;IAEvB,MAAM,KAAK,GAAG,OAAO,EAAE,KAAK,IAAI,oBAAoB,CAAC;IACrD,MAAM,MAAM,GAAG,aAAa,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;IAC9C,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;IAEtB,MAAM,WAAW,GAAG;QAClB,KAAK;QACL,UAAU,EAAE,iBAAiB;QAC7B,MAAM,EAAE,uBAAuB,EAAE;QACjC,QAAQ,EAAE;YACR;gBACE,IAAI,EAAE,MAAM;gBACZ,OAAO,EAAE,sBAAsB,CAAC,QAAQ,EAAE,eAAe,EAAE,UAAU,CAAC;aACvE;SACF;KACF,CAAC;IAEF,IAAI,eAAe,GAAG,EAAE,CAAC;IACzB,IAAI,CAAC;QACH,MAAM,QAAQ,GAAG,MAAM,KAAK,CAAC,iBAAiB,EAAE;YAC9C,MAAM,EAAE,MAAM;YACd,OAAO,EAAE;gBACP,cAAc,EAAE,kBAAkB;gBAClC,WAAW,EAAE,MAAM;gBACnB,mBAAmB,EAAE,qBAAqB;aAC3C;YACD,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,WAAW,CAAC;SAClC,CAAC,CAAC;QAEH,IAAI,CAAC,QAAQ,CAAC,EAAE,EAAE,CAAC;YACjB,MAAM,OAAO,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC;YACtD,MAAM,IAAI,KAAK,CAAC,uBAAuB,QAAQ,CAAC,MAAM,KAAK,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC;QACtF,CAAC;QAED,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,IAAI,EAG/B,CAAC;QAEF,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,EAAE,YAAY,IAAI,CAAC,CAAC;QAClD,MAAM,YAAY,GAAG,IAAI,CAAC,KAAK,EAAE,aAAa,IAAI,CAAC,CAAC;QACpD,MAAM,OAAO,GACX,CAAC,WAAW,GAAG,SAAS,CAAC,GAAG,uBAAuB;YACnD,CAAC,YAAY,GAAG,SAAS,CAAC,GAAG,wBAAwB,CAAC;QAExD,eAAe,GAAG,IAAI,CAAC,OAAO;aAC3B,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC;aAC9B,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,EAAE,CAAC;aACtB,IAAI,CAAC,EAAE,CAAC,CAAC;QAEZ,MAAM,MAAM,GAAG,oBAAoB,CAAC,eAAe,CAAC,CAAC;QAErD,OAAO;YACL,OAAO,EAAE,MAAM,CAAC,OAAO,IAAI,WAAW;YACtC,SAAS,EAAE,MAAM,CAAC,SAAS,IAAI,EAAE;YACjC,iBAAiB,EAAE,MAAM,CAAC,iBAAiB,IAAI,EAAE;YACjD,OAAO;YACP,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,MAAM,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SACnE,CAAC;IACJ,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC;QAC/B,MAAM,MAAM,GAAG,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAChE,uEAAuE;QACvE,0CAA0C;QAC1C,OAAO;YACL,OAAO,EAAE,WAAW;YACpB,SAAS,EAAE,4BAA4B,MAAM,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE;YAC1E,iBAAiB,EAAE,EAAE;YACrB,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,IAAI;YACX,WAAW,EAAE,eAAe,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,IAAI,SAAS;SACxD,CAAC;IACJ,CAAC;AACH,CAAC;AAED,8EAA8E;AAC9E,wBAAwB;AACxB,8EAA8E;AAE9E;;;;;;;;;;;;;;;GAeG;AACH,MAAM,CAAC,KAAK,UAAU,sBAAsB,CAC1C,QAAsB,EACtB,UAAgC,EAAE;IAElC,MAAM,EAAE,YAAY,GAAG,KAAK,EAAE,aAAa,EAAE,GAAG,YAAY,EAAE,GAAG,OAAO,CAAC;IAEzE,8BAA8B;IAC9B,IAAI,CAAC,YAAY,EAAE,CAAC;QAClB,MAAM,MAAM,GAAG,MAAM,YAAY,CAAC,QAAQ,EAAE,YAAY,CAAC,CAAC;QAC1D,OAAO,EAAE,GAAG,MAAM,EAAE,cAAc,EAAE,EAAE,EAAE,gBAAgB,EAAE,CAAC,EAAE,CAAC;IAChE,CAAC;IAED,MAAM,UAAU,GAAG,aAAa,EAAE,UAAU,IAAI,CAAC,CAAC;IAClD,MAAM,cAAc,GAAoB,EAAE,CAAC;IAC3C,IAAI,gBAAgB,GAAG,CAAC,CAAC;IAEzB,mBAAmB;IACnB,IAAI,WAAW,GAAG,MAAM,YAAY,CAAC,QAAQ,EAAE,YAAY,CAAC,CAAC;IAE7D,6DAA6D;IAC7D,IAAI,WAAW,CAAC,WAAW,IAAI,IAAI,EAAE,CAAC;QACpC,OAAO,EAAE,GAAG,WAAW,EAAE,cAAc,EAAE,gBAAgB,EAAE,CAAC;IAC9D,CAAC;IAED,gEAAgE;IAChE,2EAA2E;IAC3E,uEAAuE;IACvE,MAAM,cAAc,GAAG,CAAC,MAAuB,EAAE,EAAE,CAAC,CAAC;QACnD,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,KAAK,EAAE,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,eAAe,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,CAAC;YACpE,IAAI;YACJ,MAAM,EAAE,UAAU,KAAK,UAAU;SAClC,CAAC,CAAC;KACJ,CAAC,CAAC;IAEH,qBAAqB;IACrB,IAAI,OAAO,GAAG,MAAM,YAAY,CAC9B,QAAQ,EACR,WAAW,CAAC,WAAW,EACvB,cAAc,CAAC,WAAW,CAAC,EAC3B,aAAa,CACd,CAAC;IACF,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAE7B,wBAAwB;IACxB,OAAO,OAAO,CAAC,OAAO,KAAK,MAAM,IAAI,gBAAgB,GAAG,UAAU,EAAE,CAAC;QACnE,gBAAgB,IAAI,CAAC,CAAC;QAEtB,sEAAsE;QACtE,0EAA0E;QAC1E,wEAAwE;QACxE,0EAA0E;QAC1E,MAAM,YAAY,GAAmD;YACnE,GAAG,YAAY;YACf,cAAc,EACZ,oBAAoB,WAAW,CAAC,WAAW,uCAAuC;gBAClF,GAAG,OAAO,CAAC,SAAS,EAAE;gBACtB,CAAC,OAAO,CAAC,iBAAiB;oBACxB,CAAC,CAAC,wBAAwB,OAAO,CAAC,iBAAiB,EAAE;oBACrD,CAAC,CAAC,EAAE,CAAC;SACV,CAAC;QAEF,WAAW,GAAG,MAAM,YAAY,CAAC,QAAQ,EAAE,YAAY,CAAC,CAAC;QAEzD,IAAI,WAAW,CAAC,WAAW,IAAI,IAAI,EAAE,CAAC;YACpC,0CAA0C;YAC1C,MAAM;QACR,CAAC;QAED,OAAO,GAAG,MAAM,YAAY,CAC1B,QAAQ,EACR,WAAW,CAAC,WAAW,EACvB,cAAc,CAAC,WAAW,CAAC,EAC3B,aAAa,CACd,CAAC;QACF,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC/B,CAAC;IAED,OAAO,EAAE,GAAG,WAAW,EAAE,cAAc,EAAE,gBAAgB,EAAE,CAAC;AAC9D,CAAC"}

package/dist/src/benchmarks/gaia-critic.smoke.d.ts

@@ -0,0 +1,21 @@
+/**
+ * GAIA Critic Smoke Tests — ADR-135 Track D
+ *
+ * Tests for the adversarial critic agent in gaia-critic.ts.
+ * All tests use mocked responses — NO live API calls.
+ *
+ * Test coverage:
+ *  1. Critic returns "pass" → no retry, immediately returns candidate
+ *  2. Critic returns "fail" with suggestedRevision → triggers one retry
+ *  3. Critic returns "fail" twice → retries exhausted, returns last candidate
+ *  4. Critic returns "uncertain" → treated as "pass", no retry
+ *  5. API error in critic → graceful fallback, returns candidate as-is
+ *  6. Malformed JSON from critic → fallback parser extracts verdict
+ *
+ * Usage (no API key required):
+ *   npx tsx src/benchmarks/gaia-critic.smoke.ts
+ *
+ * Refs: ADR-135, #2156
+ */
+export {};
+//# sourceMappingURL=gaia-critic.smoke.d.ts.map

package/dist/src/benchmarks/gaia-critic.smoke.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"gaia-critic.smoke.d.ts","sourceRoot":"","sources":["../../../src/benchmarks/gaia-critic.smoke.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG"}