- Home
- Skills
- Agents & Orchestration
- workflow-run — 项目级 Agent Skills
workflow-run — 项目级 Agent Skills
A 9-skill pack for project-aware software delivery: dev workflow, code review, change-impact analysis, safe database and Nacos reads, and technical/release documentation.
Free
Works with the AI Agents you already use
workflow-run — 项目级 Agent Skills
Example session with this skill installed
/dev --bug 导出订单时重复返回数据
请定位原因并修复。完成后运行当前分支相对 main 的结构审查,列出受影响的接口和测试,并按项目技术方案模板生成技术方案。
- Read your context and instructions
- Compiled the workflow-run
- Generated the data export
Summary
已定位重复导出的来源:分页查询在外层重试后再次执行,导致同一个 orderId 进入结果集两次。修复采用“查询去重 + 幂等重试”两层保护,并补充了覆盖重试路径的回归测试。
根因
| Evidence | Finding | Impact |
|---|---|---|
OrderExportService.java:184 | retry wrapper re-enters the same page query | duplicate rows |
OrderMapper.xml:72 | SQL has no stable tie-breaker in ORDER BY | pages can overlap |
OrderExportServiceTest.java | existing test only covers the happy path | retry regression missing |
Implemented fix
- Kept the database query paginated, but added a deterministic tie-breaker on
created_at, id. - Deduplicated exported rows by
orderIdbefore writing the file. - Made the export retry path reuse the existing page cursor instead of restarting the page.
- Added a retry regression test and an integration assertion that every order appears exactly once.
Verification
npm test -- order-export— passed.POST /api/orders/exportmanual HTTP test — 250 exported rows, 250 unique IDs.- Branch impact review found no additional HTTP entry points or MQ consumers.
project-reviewfound no blocking issues; one follow-up note recommends observing export duration on large date ranges.
Generated technical plan
The technical plan documents the cursor change, observable behavior, test evidence, and rollback path. It uses the project's existing template rather than a generic structure.
Next steps
- Review the staged diff.
- Confirm the production date-range limit.
- Run the release checklist before deployment.
workflow-run-agent-skills.csv
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Workflow Run — Project Agent Skills
Workflow Run is a coordinated pack of nine skills for Codex, Claude Code, Qwen Code, WorkBuddy, and other agents that support the open SKILL.md standard. It gives the agent a repeatable way to inspect a project, understand the affected code, work through an explicitly requested development flow, review changes, and produce delivery documentation.
What is included
- workflow — entry point for project status, profiles, reports, PR/MR descriptions, and templates.
- dev — explicitly requested feature or bug-fix workflow, from requirements to verification and delivery.
- project-review — structured code review with field-mapping, concurrency, null-safety, resource, and security checks.
- code-graph — read-only impact analysis using an existing CodeGraph index.
- database-query — named, read-only SQL connections for data verification.
- nacos-config — read-only lookup of remote Nacos parameters and feature flags.
- create-tech-md-doc / create-release-md-doc — technical design and release plans generated from branch diffs and project templates.
- md-to-wiki — publish reviewed Markdown to an existing Confluence page.
How it works
Install the pack, then ask the agent to run /workflow status. The entry skill inspects the project and selects only the capabilities relevant to the request. Normal edits do not automatically trigger the full development flow; users opt in with /dev or by explicitly requesting it. Integrations are read-only unless a dedicated workflow asks for a write.
Best for
Java/Spring and polyglot codebases where changes need impact analysis, database or configuration evidence, consistent reviews, and documentation that follows the project's own templates.
Requirements and boundaries
Requires Node.js 22.13+ and npm. DocNexus is installed by the setup command. CodeGraph indexing, database connections, Nacos, and Confluence access are optional and must already be configured in the target project. Static graph analysis cannot see reflection, AOP routing, or every configuration-driven path.
CodeGraph must already be installed and indexed. CodeGraph cannot see AOP routing, reflection, or every config-driven path. Database, Nacos and Confluence integrations require project-specific access. The pack is optimized for Java/Spring but supports other stacks through Git and Node.js scripts.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 5 days ago
- Free with an Agensi account
Requires
Permissions
Read/write access is used only for the selected project and requested workflow artifacts. Database and Nacos skills are read-only; network and credentials are needed only for the optional integrations you enable.