[CRITICAL] Fix Agent Dispatch Pipeline — defaultDispatch Always Throws, All Tasks Fail

[devJaja]

🎯 Objective

CRITICAL / URGENT — Fix the complete agent dispatch pipeline failure. The coordinator's defaultDispatch always throws, meaning every task submitted fails immediately during execution. The entire multi-agent coordination system is non-functional.

---

📁 Files to Modify

Action	File Path	Description
Modify	backend/src/api/app.ts	Replace defaultDispatch with HTTP dispatch
Modify	backend/src/coordinator/coordinator.ts	Accept dispatch function parameter

📁 Files to Create

Action	File Path	Description
Create	backend/src/coordinator/dispatch.ts	HTTP dispatch with retries, timeouts, agent lookup

📁 Files to Modify (Tests)

Action	File Path	Description
Modify	backend/tests/e2e/pipeline.test.ts	Add real HTTP dispatch test
Create	backend/src/coordinator/dispatch.test.ts	Unit tests for dispatch

---

🔍 Root Cause Analysis

// backend/src/api/app.ts — line 397
const defaultDispatch = async (nodeId: string, task: any) => {
  throw new Error('No dispatch implementation configured'); // ← ALWAYS THROWS
};

The coordinator calls dispatch() for each DAG node. Since defaultDispatch always throws:

1. Task is created in database ✅
2. DAG is decomposed ✅
3. Coordinator starts executing nodes ✅
4. dispatch() is called → throws immediately ❌
5. All nodes marked as failed ❌
6. Task status becomes failed ❌

---

💥 Impact (Severity: CRITICAL)

• 100% of submitted tasks fail during execution
• The entire agent coordination pipeline is dead
• Research, Risk, Coding, Design, Report agents are never invoked
• Payment layer never triggers (no nodes complete)
• Frontend shows all tasks as "failed" immediately after submission
• The core product value proposition is broken

---

✅ Expected Implementation

Create backend/src/coordinator/dispatch.ts

import { SubTask } from '../types/task';
import { getAgentRegistry } from '../agents/registry';

interface DispatchOptions {
  timeoutMs: number;
  maxRetries: number;
  retryDelayMs: number;
}

const DEFAULT_OPTIONS: DispatchOptions = {
  timeoutMs: 60_000,
  maxRetries: 3,
  retryDelayMs: 1_000,
};

export async function httpDispatch(
  subtask: SubTask,
  options: DispatchOptions = DEFAULT_OPTIONS
): Promise<any> {
  const registry = getAgentRegistry();
  const agent = registry.lookupAgent(subtask.type);

  if (!agent) {
    throw new Error(`No agent available for type: ${subtask.type}`);
  }

  let lastError: Error | null = null;

  for (let attempt = 0; attempt <= options.maxRetries; attempt++) {
    try {
      const controller = new AbortController();
      const timeout = setTimeout(() => controller.abort(), options.timeoutMs);

      const response = await fetch(agent.endpoint, {
        method: 'POST',
        headers: { 'Content-Type': 'application/json' },
        body: JSON.stringify(subtask),
        signal: controller.signal,
      });

      clearTimeout(timeout);

      if (!response.ok) {
        throw new Error(`Agent ${agent.id} returned ${response.status}`);
      }

      return await response.json();
    } catch (err) {
      lastError = err as Error;
      if (attempt < options.maxRetries) {
        await new Promise(r => setTimeout(r, options.retryDelayMs * Math.pow(2, attempt)));
      }
    }
  }

  throw lastError;
}

Update app.ts

import { httpDispatch } from '../coordinator/dispatch';

// Replace defaultDispatch
const dispatch = async (nodeId: string, task: SubTask) => {
  return httpDispatch(task);
};

---

📁 Reference Files

File Path	Lines	Issue
backend/src/api/app.ts	397	defaultDispatch always throws
backend/src/coordinator/coordinator.ts	~50	Where dispatch is called
backend/src/agents/registry.ts	—	Agent endpoint lookup
backend/src/types/task.ts	—	SubTask type
backend/src/services/venice/client.ts	—	Retry pattern reference

---

✅ Acceptance Criteria

• Create backend/src/coordinator/dispatch.ts
• Implement HTTP POST to agent endpoints
• Implement configurable timeout (default 60s)
• Implement exponential backoff retry (3 attempts)
• Handle agent unavailable (503), timeout, network errors
• Replace defaultDispatch in app.ts
• Create backend/src/coordinator/dispatch.test.ts
• Add test: successful dispatch returns response
• Add test: timeout triggers retry
• Add test: agent not found throws clear error
• Update e2e pipeline test with mock HTTP server
• Verify tasks complete successfully end-to-end

[Songu3020]

Hi! I’m a full-stack developer with a strong blockchain background. I’d love to handle this issue and can deliver a clean, reliable solution. Kindly assign it to me.

[malaysiaonelove]

Hi maintain,I will love to work on this issue,I have an experience and I have already reviewed the issue please assigned me to it and I will give a good result.

[grantfox-oss]

🦊 GrantFox — @malaysiaonelove has been assigned to this issue as part of the Official Campaign | FWC26 campaign!

Next steps:

1. Open a Pull Request referencing this issue (e.g., Closes #173)
2. Your PR will be reviewed by the Epta-Node maintainers

Good luck! Track your progress on GrantFox.

[malaysiaonelove]

Resolved by PR #190 (commit cb4c1c6 'feat(backend): implement real HTTP dispatch for agent endpoints') which is already on main. The 'defaultDispatch always throws' bug no longer reproduces — the coordinator now uses makeHttpDispatch(registry) (backend/src/api/app.ts), backed by backend/src/coordinator/dispatch.ts with 60s timeout + 3 retries + exponential backoff. Verified locally: 224/226 backend tests pass including all 13 dispatch unit tests and the 10-case e2e pipeline test (including the 'HTTP dispatch integration (mock agent server)' suite).

[malaysiaonelove]

Resolved by PR #190 (commit cb4c1c6).