Prompt Library
BLACK, GREEN, BLUE, YELLOW, MATRIX, WHITE, RED, and ORANGE — eight terminal agents with dedicated domains, tool access, and project assignments, operating your platform via Claude Code.
Overview
The Terminal Agents Library is the operating crew behind Windfield IO — eight named agents, each with a dedicated terminal, an owned domain, assigned projects, and scoped tool access. BLACK, GREEN, BLUE, YELLOW, MATRIX, WHITE, RED, and ORANGE don't overlap or collide; they coordinate through a shared execution protocol that lets work route to the right specialist and hand off cleanly. This is the system prompt set and protocol that makes a multi-agent operation actually run.
Stop running one overloaded assistant. Deploy eight specialists with clear lanes and a protocol that coordinates them.
How It Works
Three phases turn a task into coordinated work across eight specialized agents.
A task is matched to the agent whose domain owns it.
Each agent works with its own tools and project scope, no collisions.
The execution protocol passes results between agents and back to you.
Column Prompts
Each column prompt runs automatically, transforming your raw input into structured, publication-ready output.
Defines an agent's domain, voice, and ownership.
Scopes which tools each agent is allowed to use.
Routes projects to the agent that owns them.
Sets the rules for how agents coordinate and hand off.
Writes the system prompt that drives each agent.
Defines how each agent reports status and health.
What You Get
Every database is pre-built, pre-connected, and ready to duplicate into your Notion workspace.
The eight agents with domains and current status.
Tools mapped to the agents permitted to use them.
Projects routed to their owning agent.
Coordination and handoff rules across agents.
The driving system prompt for each agent.
Core Concepts
Understand the foundational terms that power every prompt in this library.
The Framework
Six principles that make eight agents an operation, not chaos.
Each agent owns a domain; none overlap.
Agents touch only the tools in their lane.
Tasks land with the owning specialist automatically.
Coordination follows explicit rules, not improvisation.
Every agent reports health and status.
Add or retire an agent without breaking the rest.
Who Is This For
Whether you're a solo operator or leading a team, this system adapts to your workflow.
Run a platform with a coordinated agent crew.
Scale execution beyond one overloaded assistant.
Delegate domains to specialized agents.
Build a multi-agent system with real structure.
Model a crew of agents with clear ownership.
Move from one chatbot to an operating system.
Deep Dive
Every column prompt is engineered with specific output types and structured reasoning layers.
Generates a complete agent profile — domain, responsibilities, and operating personality — for each of the eight.
Defines the precise tool permissions per agent so each operates only within its lane.
Maps projects and task types to the owning agent so work lands with the right specialist automatically.
Defines the coordination protocol — handoffs, escalation, reporting — that keeps eight agents in sync.
Produces a tuned system prompt per agent, encoding its domain, tools, and protocol obligations.
Specifies the heartbeat and status reporting each agent emits so the operation stays observable.
A single brilliant prompt is a party trick. A coordinated system of prompts that read the same context, share the same conventions, and hand off cleanly is an operation. This library is built on the second idea: every prompt is a column in a database, every column knows what came before it, and the whole thing runs from one input instead of a dozen disconnected chat windows.
Getting Started
From purchase to first output in under 10 minutes.
One click clones every database, view, and column prompt into your own workspace.
Answer the questionnaire once — it becomes the shared memory for the whole library.
Trigger the column prompts and watch structured output populate in seconds.
Video Walkthrough
Watch how the system transforms raw input into structured output.
Product Showcase
AI-Powered
Every prompt ships with an embedded assistant trained on the library's own conventions. Ask it to adapt a prompt, explain an output, or chain two columns together — it answers in the context of your workspace, not from a blank slate.
Template Gallery
Every template is designed for immediate use. Duplicate, customize, and deploy.
The Difference
| # | Without This System | With This System |
|---|---|---|
| 01 | One overloaded assistant | Eight specialized agents |
| 02 | Agents stepping on each other | Clear, non-overlapping domains |
| 03 | Every agent can touch everything | Scoped tool access per agent |
| 04 | Manual task routing | Work routed to the owning agent |
| 05 | Coordination by improvisation | A defined execution protocol |
| 06 | No visibility into agents | Health and status reporting |
| 07 | Rigid, hard to extend | Composable — add or retire agents |
| 08 | A chatbot | An operating system |
FAQ
Everything you need to know before getting started.
Powered By
Every feature is built natively inside Notion, enhanced by AI column prompts.
Use Cases
Run a whole platform with a coordinated crew.
Hand each function to its specialist agent.
Stand up a structured agent team for a codebase.
Route tasks to owning agents automatically.
Author tuned system prompts per role.
Track agent health across the crew.
Plans & Pricing
Recurring access to the full library, its databases, and every future update. Start free, upgrade anytime.
