What does this skill do, and when should you use it?
This skill guides SkillHub contributors through the correct development sequence by defining staged workflows for local development, testing, staging validation, and pull request creation. It provides make commands to start the full stack, run tests at multiple layers, validate the containerized staging environment, and create pull requests with conventional commit messages. The skill also supports parallel agent development with isolated worktrees for feature branches.
It performs operations such as: starting the full stack (Postgres, Redis, MinIO, scanner, backend, frontend) via make dev-all, restarting the backend via make dev-server-restart, running backend unit tests via make test-backend, frontend Vitest tests via make test-frontend, Playwright E2E tests via make test-e2e-frontend, type-checking via make typecheck-web, building Docker images and frontend static files plus smoke tests via make staging, creating pull requests via make pr, and managing parallel worktrees via make parallel-init.
- A SkillHub contributor editing Java or frontend code locally wants fast iteration, so they run `make dev-all` to start all services and use hot reload or restart to see changes.
- A developer wants to verify changes before pushing, so they run `make test-backend-app` for unit tests, `make test-frontend` for frontend tests, and `make typecheck-web` for type checking.
- A developer wants to validate the containerized deployment before merge, so they run `make staging` to build images, start staging, and run smoke tests.
- A developer ready to merge a feature uses `make pr` to push the branch and create a pull request with conventional commit style.
- An agent team working on multiple features in parallel uses `make parallel-init TASK=feature-name` to create isolated worktrees.
How do you install this skill?
- Commands in skill may fail due to environment differences, users need to troubleshoot themselves.
- Documentation lacks versioning and changelog, maintenance responsibility unclear.
- No rollback or undo steps provided, manual handling needed on failure.
- Shell / CLI
- Network access
- Local filesystem
Java 21+Maven wrapperNode.js + pnpmDocker + docker composegh CLIcurl
Since this skill is part of the SkillHub repository, clone the repository and navigate to the skill directory defined by SKILL.md. The directory natively conforms to the Agent Skills standard; you can copy or link the .agents/skills/dev-workflow folder to your agent's skills directory of choice.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/iflytek/skillhub.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/.agents/skills/dev-workflow" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Follow the dev-workflow skill: start from Stage 1 and run `make dev-all`.
Trigger this skill from the repository root, for example: "Follow the dev-workflow skill: start from Stage 1 and run make dev-all." It expects prerequisites Java 21+, Maven wrapper, Node.js and pnpm, Docker, and gh CLI. Follow the four stages described in the skill: local development, testing, staging regression, and pull request.
What are this skill's strengths and limitations?
- Comprehensive staged workflow covering the entire development cycle
- Clear make commands for easy execution
- Covers end-to-end tests (Playwright) and smoke test scripts
- Supports parallel agent development with worktrees
- Conventional commit style promotes clean git history
- Tied to the SkillHub repository structure; not reusable in other projects
- Requires full local stack (Java, Node.js, Docker), heavy setup
- No automated test suite for the skill itself documented
- Make commands assume Unix environment; Windows users may need adjustments
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| SkillHub Development Workflow this page | 59 · Recommended | ★ 5.2k | 3d ago | Apache-2.0 |
| Playwright Development Guide ✓ Microsoft · Official | 45 · Not recommended | ★ 97k | 3d ago | Apache-2.0 |
| OpenCLI Adapter Authoring Guide | 56 · Use with care | ★ 30k | 17d ago | Apache-2.0 |
| ClawTeam Repository Development Skill | 53 · Use with care | ★ 5.5k | 5mo ago | MIT |
| HSB QA Test Runner ✓ NVIDIA · Official | 46 · Use with care | ★ 3.5k | 3d ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill only guides development workflows, no external side effects or sensitive data handling, low permission needs. But no explicit confirmation steps or data-flow transparency, no rollback mechanism. Publisher unverified, trust based on documentation completeness only, no red flags.
Commands and steps are consistent within the documentation, but no tests or error handling details, relies on external tools (Docker, Java), key paths not reproducible in static review. Score based on documentation self-consistency only.
Clear use cases (local dev, testing, PR), but boundaries and trigger conditions not explicitly defined, may not prevent false triggering. No Chinese support mention, but deployment docs mention China mirrors, environment fit partially satisfied.
Well-structured docs with prerequisites, common commands, troubleshooting, but missing versioning, changelog, maintenance responsibility. Apache-2.0 license present but update path unclear.
Skill can guide through development workflow, but user still needs manual verification, output not automated. Core task achieved, marginal value depends on doc quality, no actual verification.
No test results or third-party execution evidence, only docs and repo CI, but skill itself not covered by tests. Key claims not independently verifiable.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →