AI Agent Hub
Back to skills
Dev Flow Pro Development Workflow icon

Dev Flow Pro Development Workflow

Development Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Follow https://skillhub.cn/install/skillhub.md to install @user_8e4c5a5c/dev-flow-pro.

About this skill

Problem

When a coding agent is asked to modify a complex repository directly, it may skip planning, patch code before adding tests, omit review, or lose progress when context expires. dev-flow turns these risks into explicit workflow constraints, requiring scenario routing, pre-checks, and stage-level traceability.

How It Works

It routes tasks into A new feature, B bug fix, C rework, D refactor, E assessment, F hotfix, or G dependency upgrade, then loads the matching flow file. Execution enforces rules such as no coding without a plan, no code changes without tests, no release without review. If optional tooling is unavailable, it falls back to manual task_plan.md, progress.md, and findings.md. It also requires updating progress before context exhaustion, isolating tests or subagent experiments in git worktree, and handling repeated failures, blocked test suites, or missing plan details.

Boundaries

Use it for multi-step, multi-file engineering changes that need review and recovery. It is not intended for one-line fixes, configuration changes, or copy edits. If the repository lacks a passing test baseline, installed dependencies, or is not at the repo root, fix those prerequisites first.

Use Cases

  • Receive a login audit feature request, then draft a plan, add failing tests, implement, and review.
  • Debug intermittent production 500 errors by reproducing the failure, finding root cause, and patching with tests.
  • Refactor order settlement while keeping the public API stable, with a verified test baseline.
  • Continue multi-file work across sessions using task_plan.md and progress.md to restore state.

Best For

  • Engineering leads directing coding agents on multi-file changes who need planned, tested, reviewed work.
  • Backend engineers handling production incidents who need reproduction and root-cause fixes.
  • Architects refactoring internal modules while keeping public APIs stable and tests green.
  • Developers using agents across sessions who need progress files to restore context.