Plan to Project

Reference: See things3 skill for task requirements, tag taxonomy, and MCP operations.

Convert a plan, specification, or research session into a well-structured Things 3 project with tasks that agents can pick up and execute autonomously.

When to Use

Use this skill when:

• You've completed research or planning and want to capture next steps
• You have a plan file (from plan mode) to convert
• The user describes a project and wants it tracked in Things 3
• Breaking down a large initiative into actionable tasks

Arguments

The skill receives a project name or plan file path as $ARGUMENTS. Examples:

• /plan-to-project Build things CLI - Create from conversation context
• /plan-to-project ~/.claude/plans/my-plan.md - Create from plan file

Workflow

1. Gather Context

If a plan file is provided, read it:

Read(file_path="$ARGUMENTS")

If creating from conversation, synthesize the key decisions, architecture, and tasks from the discussion.

2. Create Project with Rich Context

Create the project with comprehensive notes that give agents full context:

mcp__things__add_project(
    title="Project Title",
    notes="## Overview
[1-2 sentence summary of what this project accomplishes]

## Why This Matters
[Business/personal value - why is this worth doing?]

## Architecture/Approach
[Key technical or strategic decisions made]

## Key Resources
- [Links to docs, repos, references]
- [Related files or tools]

## Context from Planning
[Important decisions, constraints, or findings from research]",
    area_title="[Appropriate Area]",
    tags=["1h"]  # Estimated total time
)

3. Break Down Into Agent-Ready Tasks

For each task, create with:

1. Verb-first title - Action-oriented, specific
2. Detailed notes - Everything an agent needs to execute independently
3. Time estimate tag - 2m, 5m, 25m, or 1h+
4. Agent/Queued tag - Ready for agent pickup

Task Note Template

mcp__things__add_todo(
    title="Implement [specific thing]",
    notes="## Context
[Why this task exists, what problem it solves]

## Steps
1. [Specific step 1]
2. [Specific step 2]
3. [Specific step 3]

## Implementation Details
[Technical specifics, code patterns, file locations]

## Acceptance Criteria
- [How to know it's done]
- [What to verify]

## References
- [Links, docs, examples]",
    list_id="<project-uuid>",
    tags=["25m", "Agent/Queued"]
)

4. Task Ordering

Create tasks in logical execution order:

1. Foundation first - Setup, infrastructure, dependencies
2. Core functionality - Main features in order of dependency
3. Polish last - Tests, docs, distribution

Things 3 preserves creation order, so agents will work through tasks in the right sequence.

5. Task Sizing Guidelines

If a task is larger than 1h+, consider:

• Breaking into subtasks within the same project
• Creating a sub-project

6. Report Creation Summary

After creating the project, report:

Created project: [Project Name]

Tasks ([count] total):

Phase 1: [Phase Name]
1. [ ] Task 1 [25m]
2. [ ] Task 2 [25m]

Phase 2: [Phase Name]
3. [ ] Task 3 [1h]
...

All tasks tagged Agent/Queued and ready for pickup.

Quality Checklist

Before finishing, verify each task has:

• Verb-first title - Starts with action verb
• Context section - Why this task exists
• Steps or implementation details - How to do it
• Acceptance criteria - How to know it's done
• Time estimate tag - 2m/5m/25m/1h+
• Agent/Queued tag - Ready for pickup
• Project assignment - Belongs to the created project

Example: From Research to Project

Input: A conversation researching "how to build a Go CLI for Things 3"

Output:

Project: Build `things` Go CLI for Things 3
Notes:
## Overview
High-performance CLI using SQLite reads + URL scheme writes.

## Why This Matters
- 10-100x faster than AppleScript
- Enables field selection to reduce tokens for Claude
- No existing tool combines SQLite + URL writes

## Architecture
cmd/ - Cobra commands
internal/db/ - SQLite queries
internal/urlscheme/ - URL scheme writes
internal/output/ - JSON, table formatters

## Key Resources
- things.py: https://github.com/thingsapi/things.py
- URL Scheme: https://culturedcode.com/things/support/articles/2803573/

Tasks (20):
1. Initialize Go module and create repo structure [25m]
2. Implement database connection with auto-discovery [25m]
3. Port Things date format conversion [25m]
...

Anti-Patterns

DON'T create vague tasks

Bad: "Work on database stuff" Good: "Implement database connection with auto-discovery"

DON'T skip context in notes

Agents need full context to work autonomously. Include why, how, and acceptance criteria.

DON'T create tasks without time estimates

Every task needs a time tag for prioritization.

DON'T forget Agent/Queued tag

Tasks without this tag won't be picked up by agents.

DON'T create orphan tasks

Every task must belong to the project you just created.

MCP Tools Reference