incremental-planninglisted
Install: claude install-skill Accelerated-Innovation/governed-ai-delivery
# Incremental Planning
## Purpose
Plan implementation work as a sequence of the smallest independently
demonstrable increments.
This skill **does not write code**. Its responsibility is to determine the next
demonstrations that should exist, not how they should be implemented.
Implementation, testing, refactoring, commits, pull requests, and releases are
responsibilities of other skills.
## Core Principle
Grow system capability by admitting one new coherent behavior at a time.
Each proposed increment should expand the system's capabilities in exactly one
coherent way while leaving the system in a deployable, demonstrable, and
coherent state.
## Planning Algorithm
Given any implementation request:
1. Classify the request.
2. Determine whether it already represents a demonstrable increment.
3. If not, discover the smallest sequence of demonstrable increments.
4. Validate each proposed increment.
5. Stop.
Do not produce implementation guidance. Do not generate code.
## Request Classification
- **Ready Increment** — already the smallest independently demonstrable
increment. → Implement as a single increment.
- **Composite Increment** — contains multiple independently demonstrable
increments. → Split into the smallest sequence of independently demonstrable
increments.
- **Hidden Increment** — phrased in implementation terms but implies one or more
demonstrable behaviors (e.g. "Implement UploadService", "Build Search API",
"Add CSV Import"). → Recover the demo