Menü
Memory-based agent communication protocol for multi-agent workflows
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
End-to-end pipeline to augment any repository for AI-assisted development. Installs opinionated engineering workflows, generates documentation, and configures Claude Code or Cursor.
Destructive command safety guardrails. Warns before dangerous operations in production and shared environments.
Transform a knowledge graph into a human-readable /docs folder with markdown documentation that both humans and agentic workflows can reference.
Systematic root-cause debugging and incident investigation. No fixes without understanding the problem first.
Architecture review before implementation. Walk through design, failure modes, scope, and test strategy before writing code.
Engineering retrospective and velocity analytics from git history. Generates weekly insights on team productivity, code quality, and contribution patterns.
| name | agent-coordination |
| version | 1.0.0 |
| description | Memory-based agent communication protocol for multi-agent workflows |
| author | iscmga |
| tags | ["agents","coordination","multi-agent","subagents"] |
| triggers | {"keywords":["subagent","multi-agent","coordinate","delegate","parallel","orchestrate"]} |
Agents coordinate through shared memory files, not direct messaging. Each agent reads inputs from and writes outputs to well-known paths. This works natively in Claude Code (via the Agent tool + workspace files) and Cursor (via workspace files).
memory/
runtime/
coordinator.md # Current task decomposition and status
handoffs/
{agent-role}.md # Output from each agent role
decisions.md # Key decisions made during coordination
The coordinator (main Claude Code session) decomposes work and writes:
# Task: [description]
## Subtasks
- [ ] research: [what to investigate]
- [ ] implement: [what to build]
- [ ] test: [what to verify]
- [ ] review: [what to check]
## Constraints
- [shared context all agents need]
## Status: in-progress
Each subagent writes its output to memory/runtime/handoffs/{role}.md:
# Research Output
## Findings
- [key finding 1]
- [key finding 2]
## Recommendations
- [recommendation]
## Files Touched
- path/to/file.ts:42
## Status: complete
The coordinator reads all handoff files, integrates results, and updates coordinator.md status.
Use the Agent tool to spawn subagents. Each subagent gets a focused prompt:
Subagent prompt pattern:
"Read memory/runtime/coordinator.md for context.
Do [specific task].
Write your results to memory/runtime/handoffs/{role}.md.
Do not modify files outside your scope."
Spawn independent subagents in parallel. Wait for dependent ones sequentially.
In Cursor, coordination happens across chat sessions: