Multi-Agent Handoff

Coordinate planner, implementer, and reviewer agents with explicit scope and outputs.

CopilotCursorKilo
AuthorNeexoCore
Updated
multi-agentplanningreview

Install in your repo

You get: 1 SKILL.md

Copy one prompt into Cursor, Copilot, Kilo, or any agent with web access.

Install the Neexo skill "Multi-Agent Handoff" (multi-agent-handoff) in this repository.

Manifest: https://awesome.neexo.dk/api/install-manifest?ids=skills%2Fmulti-agent-handoff
Catalog: https://awesome.neexo.dk/skills/multi-agent-handoff

Steps:
1. Fetch https://awesome.neexo.dk/api/install-manifest?ids=skills%2Fmulti-agent-handoff and read the file list.
2. Fetch https://awesome.neexo.dk/api/skills/multi-agent-handoff/bundle and write every file to the correct project skill folder.
3. Detect IDE from this repo (.cursor → Cursor, .kilo → Kilo).
4. Preserve subdirectory structure (references/, agents/, scripts/, assets/).
5. Confirm which files you wrote.
Advanced — manual CLI
CLI — Cursor + Kilonpx awesome-neexo install skill multi-agent-handoff --all
CLI — Cursor onlynpx awesome-neexo install skill multi-agent-handoff --cursor
Copilot — full plugincopilot plugin install neexo-planning@awesome-neexo
Bundle API (JSON)https://awesome.neexo.dk/api/skills/multi-agent-handoff/bundle

Overview

Use multiple agents when one perspective is not enough — not by default on every task.

Workflow chain

Discovery: neexo-brainstorm → scoped-plan | plan-feature-change → implementer → reviewer

When to delegate

  • Open-ended improvement: brainstorm agent → scoped-plan or plan-feature-change → implementer
  • HIGH-risk changes: plan agent → implementer (after approval) → reviewer
  • Large features: planner produces scoped plan; worker implements one slice at a time
  • Disagreement: consensus-style second pass on security or data integrity

Handoff template

HANDOFF
From: [brainstorm|planner|implementer|reviewer]
To: [next role]
Stack: [playbook slug]
Top ideas: [from brainstorm, if applicable]
Approved files: [list]
Non-goals: [list]
Validation: [commands]
Open risks: [list]

Cost discipline

  • Each agent gets minimal context — prior agent output + relevant files only
  • Do not re-scan the full repository on handoff
  • Prefer one skilled agent over three shallow passes for LOW tasks

Raw content

Copy into your project — e.g. .instructions.md, .agent.md, or SKILL.md

## Overview

Use multiple agents when one perspective is not enough — not by default on every task.

## Workflow chain

```text
Discovery: neexo-brainstorm → scoped-plan | plan-feature-change → implementer → reviewer
```

## When to delegate

- Open-ended improvement: brainstorm agent → scoped-plan or plan-feature-change → implementer
- HIGH-risk changes: plan agent → implementer (after approval) → reviewer
- Large features: planner produces scoped plan; worker implements one slice at a time
- Disagreement: consensus-style second pass on security or data integrity

## Handoff template

```text
HANDOFF
From: [brainstorm|planner|implementer|reviewer]
To: [next role]
Stack: [playbook slug]
Top ideas: [from brainstorm, if applicable]
Approved files: [list]
Non-goals: [list]
Validation: [commands]
Open risks: [list]
```

## Cost discipline

- Each agent gets minimal context — prior agent output + relevant files only
- Do not re-scan the full repository on handoff
- Prefer one skilled agent over three shallow passes for LOW tasks

Next steps

Matched by shared tags and category. Browse all plugins