kkirikiri

Coordinate a Codex-native agent team for delegated parallel work.

114|19|Updated Apr 24, 2026

Search Tags:#planning #codex #delegation #multi-agent #shared-memory #kkirikiri

Ready to run

npx skills add https://github.com/fivetaku/gptaku-plugins-codex --skill kkirikiri

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: kkirikiri
Source: https://github.com/fivetaku/gptaku-plugins-codex/tree/main/plugins/kkirikkiri-codex/skills/kkirikkiri
Command: npx skills add https://github.com/fivetaku/gptaku-plugins-codex --skill kkirikiri

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Assembles and coordinates a Codex-native agent team from a plain-language request, enabling delegation, parallel work, and collaborative execution within a single Codex session.

Core Features & Use Cases

Bounded team orchestration: spawns a lead plus a small set of bounded sub-agents to tackle research, development, review, and coordination tasks.
Shared memory and planning: uses a centralized memory store (.kkirikkiri) to track TEAM_PLAN, TEAM_PROGRESS, and TEAM_FINDINGS, ensuring continuity across agents.
Guided prompts and references: reads from references to guide interview workflows, presets, and team prompts for consistent execution.
Guardrails and safety: protects against scope creep by requiring explicit delegation requests and clear acceptance criteria.

Quick Start

Provide your plain-language request and I will assemble a Codex-native agent team to execute it.

Frequently Asked Questions about kkirikiri

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

Does multi-agent delegation work for tasks requiring research, planning, and execution together?▼

Multi-agent delegation applies to tasks requiring research, planning, and execution together. The lead-driven workflow spawns bounded sub-agents to tackle these distinct phases concurrently within a single session.

How do I coordinate multi-agent teams for parallel task execution?▼

Multi-agent teams are coordinated by spawning a lead agent and bounded sub-agents within a single session. This orchestration allocates ownership across agents and captures findings in a shared memory store for parallel work.

Can I use a shared memory store to maintain agent progress during delegation?▼

A centralized shared memory store maintains agent progress during delegation by tracking team plans, progress, and findings. This ensures continuity across all spawned agents throughout the execution lifecycle.

What is the best way to prevent scope creep when delegating tasks to sub-agents?▼

To prevent scope creep when delegating tasks, the workflow requires explicit delegation requests and clear acceptance criteria. These guardrails protect against uncontrolled expansion of sub-agent responsibilities.

How do I assemble a development team from plain-language requests without manual setup?▼

Assembling a development team from plain-language requests is handled by reading role prompts and presets from references. This guides interview workflows and spawns agents automatically without manual configuration.

Does multi-agent delegation work for tasks requiring research, planning, and execution together?▼