todo-utilities

Todo Utilities

Purpose

This is a PATTERN LIBRARY, not a skill you invoke.

Skills read these patterns and copy them inline into their own workflows. This ensures every skill tracks progress the same way — consistent user experience, no reinventing the wheel.

Why not invoke as a Task? Because TodoWrite calls inside a Task() are invisible to the user. The whole point of progress tracking is user visibility.

Core Principle

TodoWrite = display layer (what the user sees).

Always update TodoWrite:

1. BEFORE invoking an agent → mark task in_progress
2. AFTER agent returns → mark task completed
3. On error → update activeForm to show what's happening, don't hide failures

Patterns

See: references/todo-patterns.md

The full pattern library with examples. Contains:

1. Sequential Agent Invocation — Tasks run one after another
2. Parallel Agent Invocation — Multiple independent agents
3. Error Recovery — Graceful failure with visible retry
4. Workflow Gate — User approval required before continuing
5. Long-Running Progress — Show progress counts during bulk operations

Anti-Patterns

What NOT to do. Common mistakes that break user visibility:

How Skills Reference This

Each BCOS skill that uses agent delegation or multi-step workflows should include:

### Progress Tracking

This skill follows the TodoWrite patterns from `todo-utilities`.
See `.claude/skills/todo-utilities/references/todo-patterns.md` for the full library.

Then copy the relevant pattern inline. The pattern is the source of truth — if it changes, update the patterns doc and all skills inherit the intent.