ruvnet/ruflo · error

SONA adaptation failed: ${error}

Error message

SONA adaptation failed: ${error}

What it means

Thrown by the SONA query-adaptation path in @claude-flow/neural: the whole adaptQuery body (query transform, pattern retrieval, route inference, confidence scoring) is wrapped in one try/catch, and any lower-level failure is re-thrown as this umbrella error with the original message appended after the colon. The prefix is only an annotation; the suffix is the real cause. The adaptation-time metric is only recorded on success, so this throw also means no timing was captured.

Source

Thrown at v3/@claude-flow/neural/src/sona-integration.ts:237

      const patterns = this.engine.findPatterns(
        Array.from(context.queryEmbedding),
        5
      );

      // Determine suggested route from patterns
      const suggestedRoute = this.inferRoute(patterns, context);
      const confidence = patterns.length > 0 ? patterns[0].avgQuality : 0.5;

      this.adaptationTimeMs = performance.now() - startTime;

      return {
        transformedQuery: new Float32Array(transformedQuery),
        patterns,
        suggestedRoute,
        confidence,
      };
    } catch (error) {
      throw new Error(`SONA adaptation failed: ${error}`);
    }
  }

  /**
   * Get last adaptation time
   *
   * @returns Adaptation time in milliseconds
   */
  getAdaptationTime(): number {
    return this.adaptationTimeMs;
  }

  /**
   * Get last learning time
   *
   * @returns Learning time in milliseconds
   */
  getLearningTime(): number {

View on GitHub (pinned to fa13ee4ad6)

Solutions

1. Read the text after 'SONA adaptation failed:' — it names the actual failing operation; debug that, not the wrapper
2. Ensure initialization (pattern store load, mode implementation) has completed before the first adaptQuery call
3. Validate the query up front: finite values, expected dimensionality, Float32Array
4. If the inner error indicates corrupted state, rebuild patterns from trajectories or re-serialize with the current package version
5. Reproduce with logging around the transform/pattern-retrieval steps to pinpoint the throw site

Example fix

// before
const r = await sona.adaptQuery(anyOldVector, ctx);
// after
if (!(query instanceof Float32Array) || !query.every(Number.isFinite)) {
  throw new TypeError('query must be a finite Float32Array');
}
try {
  const r = await sona.adaptQuery(query, ctx);
} catch (e) {
  throw new Error(String(e).replace('SONA adaptation failed: ', ''), { cause: e });
}

// validate the query before adaptation
const ok =
  query instanceof Float32Array &&
  query.length > 0 &&
  query.every(Number.isFinite);
if (!ok) throw new TypeError('query must be a non-empty finite Float32Array');
const r = await sona.adaptQuery(query, ctx);

function isFiniteVector(v: unknown): v is Float32Array {
  return v instanceof Float32Array && v.every(Number.isFinite);
}

try {
  const r = await sona.adaptQuery(query, ctx);
} catch (e) {
  const msg = e instanceof Error ? e.message : String(e);
  if (msg.startsWith('SONA adaptation failed:')) {
    // the suffix is the real cause — surface it, keep the original as cause
    throw new Error(msg.slice('SONA adaptation failed:'.length).trim(), { cause: e });
  }
  throw e;
}

When it happens

Trigger: Any exception raised inside the try block: calling adaptQuery before the integration/pattern store is initialized, passing a query vector with non-finite values or wrong dimensionality, corrupted or version-skewed persisted patterns making transformQuery or pattern retrieval throw, or inferRoute failing on inconsistent mode/EWC state.

Common situations: Reusing a SonaIntegration instance after the underlying neural store was closed or reloaded mid-flight; loading SONA state written by a different package version so internal shape checks fail; passing Float64Array or a plain array where a Float32Array is expected; concurrent close() during an in-flight adaptation.

Related errors

• SONA learning failed: ${error}
• Failed to execute MCP tool '${toolName}': ${error instanceof
• Unknown algorithm: ${algorithm}
• FlashAttention: Empty input arrays
• FlashAttention: Keys and values must have same count. Got ${

AI-assisted analysis of ruvnet/ruflo@fa13ee4ad6 (2026-08-18). Data as JSON: /api/errors/fec338389ff2a950. Report an issue: GitHub.