Implement Incremental (phase-loop orchestrator)

What it does

Implement Incremental provides a linear phase-loop orchestration layer for feature implementation plans. It reads a plan/ manifest, discovers phase files, and drives a controlled loop: mark a phase in_progress, delegate tasks to specialist subagents, validate output against the spec, and require user confirmation before moving on. The skill focuses on preserving human-in-the-loop checkpoints, updating phase frontmatter and plan README checkboxes, and summarizing agent outputs (files changed, tests, blockers) instead of dumping raw agent responses.

When to use it

Use this skill for medium-complexity feature work where you want clear phase boundaries, auditability, and manual approval between stages — for example, rolling out a backend feature with separate design, implementation, review, and test phases, or coordinating cross-functional agent teams where drift detection matters. It is not intended to write code directly; it delegates implementation to specialist agents.

What's included

• Scripts: none detected in the skill directory (has_scripts=false)
• References: linked output-format and perspective docs inside the repo (has_references=false in fetch metadata)
• Instructions: detailed phase loop, state/interface definitions, constraints, workflow steps for initialize, execute, validate, update status, and complete. The skill documents phase file formats and update rules for plan/README.md and frontmatter.

Compatible agents

Best suited for agentic runtimes that support subagent spawning and file edits (Claude Code, Claude-based agent frameworks, or similar orchestrators).