create-spec

name	create-spec
description	Create a spec.md file based on task.md and lesson.md files.

Create Specification for AI Agent Implementation

• Create a spec.md based in the root of the project on spec-template.md to serve as a coding guideline for the agent implementing the solution described in task.md.
• Use lesson.md as an extended reference and inspiration for the possible ways to implement the solution.
• Adhere to the AGENTS.md guidelines in the project.
• Ask follow-up questions to clarify the requirements and drive key decisions if needed.
• As a last step, make sure that the spec is not overly bloated, is easy to read and follow, but is detailed enough to guide the agentic implementation.

--- Spec.md Template ---

AI Agent for [Task Name]

1. Overview & Goal

Task Summary

Hardcoded Inputs / Initial Data

Final Deliverable

---

2. Agent Persona & Prompt Strategy

System Prompt (for the inner LLM agent)

<!-- Provide the template for the system prompt used to instruct the agent. -->

You are a [Persona Description]. Your job is to [Main Task].

## Your workflow of agent loop

1. STEP ONE: [Action]
2. STEP TWO: [Action]
3. ...

## Rules

- RULE ONE: [Constraint]
- RULE TWO: [Constraint]

---

3. Tool Definitions (Function Calls)

3.1 [tool_name]

Description:

Input Schema:

// Provide the JSON schema for the tool arguments
{
	"type": "object",
	"properties": {
		"param1": {
			"type": "string",
			"description": "..."
		}
	},
	"required": ["param1"]
}

Behavior:

Return value:

// Provide the expected JSON return structure
{
	"result": "..."
}

---

4. Execution Flow

<!-- Provide an ASCII flowchart or step-by-step logic of the expected agent execution -->
START
  │
  ├─ 1. [Initial Step]
  │
  ├─ 2. [Next Step / Sub-loop]
  │
  └─ END ([Final Action])

Key Decision Points

---

5. Dependencies & Environment

package.json additions

Package	Purpose
example-pkg	Does something

Environment Variables (.env)

# List required environment variables
API_KEY=<your api key>

Project Structure

# Suggest a basic directory layout
src/
  index.ts          # Entry point
  agent.ts          # Agent logic
  tools.ts          # Tool implementations
  prompts.ts        # Prompts

---

6. Key Implementation Notes

1. [Note 1]
2. [Note 2]

---

7. Acceptance Criteria

• Requirements met
• Tests pass
• Error handling built in