borghei/Claude-Skills / engineering/agenthub
agenthub
VerifiedRuns 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.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 9.8 KB |
| assets/sample_session.json | config | 4.5 KB |
| assets/sample_workflow.json | config | 1.5 KB |
| references/multi-agent-patterns.md | markdown | 5.6 KB |
| references/operations-and-quality.md | markdown | 2.9 KB |
| references/orchestration-core.md | markdown | 4.9 KB |
| scripts/board_manager.py | script | 10.9 KB |
| scripts/dag_analyzer.py | script | 13.5 KB |
| scripts/result_ranker.py | script | 11.0 KB |
| scripts/session_manager.py | script | 12.7 KB |
| skills/board.md | markdown | 2.2 KB |
| skills/eval.md | markdown | 2.2 KB |
| skills/init.md | markdown | 2.0 KB |
| skills/merge.md | markdown | 2.3 KB |
| skills/run.md | markdown | 2.1 KB |
| skills/spawn.md | markdown | 1.9 KB |
| skills/status.md | markdown | 2.1 KB |
references/operations-and-quality.md
View on GitHub# Best Practices, Pitfalls, Troubleshooting & Success Criteria
Read this when designing agent scopes, debugging a stuck or incoherent workflow, or validating an orchestration against the quality bar.
## Best Practices
1. **Small, focused agent scopes** -- each agent should have a single clear objective
2. **Explicit inputs/outputs** -- never rely on implicit shared state between agents
3. **Quality gates between stages** -- evaluate before passing outputs downstream
4. **Timeout per agent** -- prevent runaway agents from blocking the workflow
5. **Retry with context** -- when retrying a failed agent, include the failure reason
6. **Merge strategy documented** -- how competing or complementary outputs combine
7. **Critical path awareness** -- optimize the longest dependency chain first
8. **Idempotent agents** -- agents should produce the same output given the same input
## Common Pitfalls
| Pitfall | Why It Happens | Fix |
|---------|---------------|-----|
| Cycle in DAG | Agent A depends on B which depends on A | Run dag_analyzer.py before execution |
| Output format mismatch | Agent B expects JSON, Agent A produces markdown | Define explicit output schemas per agent |
| Single bottleneck agent | One agent depends on everything | Restructure DAG to parallelize dependencies |
| Lost context between agents | Outputs too terse for downstream use | Require structured output with context preservation |
| Quality degradation in merge | Naive concatenation loses coherence | Use a dedicated merge agent with synthesis instructions |
| Runaway execution time | No timeouts, retry loops | Set timeout_per_agent and max retries |
## Troubleshooting
| Problem | Cause | Solution |
|---------|-------|----------|
| Workflow hangs at agent N | Dependency not met or agent timeout | Check board for PENDING agents; verify upstream completed; check timeout config |
| Merged output is incoherent | No merge strategy defined | Use the merge sub-skill with explicit synthesis instructions |
| Agent produces wrong format | Input/output contract unclear | Define JSON schemas for agent inputs and outputs |
| DAG validation fails with cycle | Circular dependency in definition | Use dag_analyzer.py to identify the cycle; restructure the dependency chain |
| Quality eval fails everything | Threshold too strict for task complexity | Lower threshold or add a revision step before eval |
## Success Criteria
- **DAG validation passes** on every workflow definition before execution
- **Parallel execution utilization above 60%** -- agents running in parallel most of the time
- **Quality gate pass rate above 80%** -- agent outputs meet threshold on first attempt
- **End-to-end execution time within 2x critical path** -- parallelization delivers real speedup
- **Zero lost outputs** -- every agent's output is captured and available for merge/review
- **Merge coherence score above 0.7** -- final merged output reads as a unified deliverable
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 472k | Updated | Claude CodeCodex | Unverified | Install |
| 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: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| 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 available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |