Create and structure a new autonomous project — "/new-project <what project does>
Install with the open skills CLI (global, non-interactive — available in every Claude Code session):
npx skills add nearai/ironclaw --skill "new-project" -g -a claude-code -yOr manually — copy the SKILL.md below into:
~/.claude/skills/new-project/SKILL.md---
name: new-project
version: 0.2.0
description: Create and structure a new autonomous project — "/new-project <what project does>"
activation:
keywords:
- project
- create project
- new project
- set up project
- autonomous workspace
- campaign
- department
- company project
patterns:
- "create a (new )?project"
- "set up.*project"
- "organize.*into.*project"
- "new.project"
- "/new.project"
tags:
- project-management
- organization
- goals
max_context_tokens: 2000
---
# New Project
Create an autonomous project workspace using `memory_write` and `mission_create`.
## Step-by-step procedure
Given the user's description of what the project does, derive a short slug (lowercase, hyphens, e.g. `ai-research`). Then execute these steps **sequentially** (one tool call at a time — do NOT batch calls that depend on each other):
### 1. Write AGENTS.md
```
memory_write(target: "projects/{slug}/AGENTS.md", content: "# {Project Name}\n\n{What the agent should know about this project: domain, stakeholders, priorities, constraints, tools/APIs to use.}")
```
This file is loaded into the system prompt for every mission in this project. Make it specific and actionable.
### 2. Write context.md
```
memory_write(target: "projects/{slug}/context.md", content: "# {Project Name} — Context\n\n## Overview\n{What the project is and why it exists.}\n\n## Current State\n{What is known so far.}")
```
### 3. Write goals.md (if the project has clear goals)
```
memory_write(target: "projects/{slug}/goals.md", content: "# Goals\n\n- Goal 1\n- Goal 2\n...")
```
Include measurable targets when possible. If the project would benefit from tracked metrics, add a metrics section:
```
## Metrics
| Metric | Unit | Target | How to measure |
|--------|------|--------|----------------|
| {name} | {unit} | {target} | {evaluation instruction — tell the agent HOW to check this: API call, file to read, command to run} |
```
### 4. Create missions
Create recurring missions scoped to the project. Use the **project name** or **slug** as `project_id` (the engine resolves it to the correct project):
```
mission_create(name: "...", goal: "...", cadence: "daily", project_id: "{Project Name}")
```
Choose appropriate cadences: `hourly`, `daily`, `weekly`, `monthly`, or cron expressions like `0 9 * * 1-5`.
## Project structure convention
```
projects/
{slug}/
AGENTS.md # Agent instructions (loaded into system prompt)
context.md # Background knowledge, current state
goals.md # Goal breakdown with optional metrics
research/ # Research and analysis outputs
reports/ # Generated reports
```
## Rules
- Execute tool calls **one at a time**, sequentially. Wait for each result before the next call.
- Always pass `project_id` when creating missions. Without it, missions land in the Default project.
- AGENTS.md must be written first — it gives the agent project context.
- Keep the response concise. After setup, summarize what was created in a short list.
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
Use when you have a written implementation plan to execute in a separate session with review checkpoints