Skip to content

borghei/Claude-Skills / engineering/agenthub

agenthub

Verified

Runs a multi-agent job as a dependency graph: validates the DAG, tracks each agent's state on a board, then ranks and merges their outputs. Use when spawning parallel agents, monitoring a run, or merging results.

FileKindSize
SKILL.mdskill md9.8 KB
assets/sample_session.jsonconfig4.5 KB
assets/sample_workflow.jsonconfig1.5 KB
references/multi-agent-patterns.mdmarkdown5.6 KB
references/operations-and-quality.mdmarkdown2.9 KB
references/orchestration-core.mdmarkdown4.9 KB
scripts/board_manager.pyscript10.9 KB
scripts/dag_analyzer.pyscript13.5 KB
scripts/result_ranker.pyscript11.0 KB
scripts/session_manager.pyscript12.7 KB
skills/board.mdmarkdown2.2 KB
skills/eval.mdmarkdown2.2 KB
skills/init.mdmarkdown2.0 KB
skills/merge.mdmarkdown2.3 KB
skills/run.mdmarkdown2.1 KB
skills/spawn.mdmarkdown1.9 KB
skills/status.mdmarkdown2.1 KB

skills/init.md

View on GitHub
# Sub-Skill: Initialize Workflow

**Parent:** agenthub
**Trigger:** "create multi-agent workflow", "design agent DAG", "initialize orchestration"

## Purpose

Create a new multi-agent workflow definition. Guides the user through decomposing a complex task into agent nodes, defining dependencies, and validating the resulting DAG.

## Workflow

### Step 1: Decompose the Task

Break the overall objective into discrete sub-tasks:
- Each sub-task should be completable by a single agent in one session
- Sub-tasks should have clear inputs and outputs
- Identify which sub-tasks can run in parallel vs must be sequential

### Step 2: Define Agents

For each sub-task, define an agent node:
```json
{
  "id": "agent-name",
  "task": "Clear description of what this agent does",
  "inputs": ["list of required inputs"],
  "outputs": ["list of produced outputs"],
  "dependencies": ["ids of agents that must complete first"],
  "config": {
    "timeout": 300,
    "retries": 1,
    "quality_threshold": 0.7
  }
}
```

### Step 3: Build Dependency Graph

Map the edges:
- Root nodes: agents with no dependencies (start first)
- Terminal nodes: agents with no dependents (feed into merge)
- Ensure every output needed by a downstream agent is produced by an upstream agent

### Step 4: Validate

Run `dag_analyzer.py` on the definition:
```bash
python scripts/dag_analyzer.py --workflow workflow.json --validate
```

Checks:
- No cycles in the graph
- All input references resolve to upstream outputs
- No orphaned nodes (unreachable from root or terminal)
- Critical path length is reasonable

### Step 5: Save Definition

Write the validated workflow definition to a JSON file for execution.

## Inputs

| Input | Required | Description |
|-------|----------|-------------|
| Task description | Yes | The overall objective to decompose |
| Max parallel | No | Maximum concurrent agents (default: 3) |
| Timeout | No | Per-agent timeout in seconds (default: 300) |

## Outputs

- Validated workflow definition (JSON)
- DAG visualization (text-based)
- Critical path analysis
- Estimated execution time
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexUnverifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall