xrk-ai-workflow
Skill140 starsChanged 8 months agoMITView on GitHub
当你需要开发/调试 AiWorkflow 工作流、出站裁剪、策略安全、RAG 上下文增强、MCP 工具作用域时使用。
In this repository43 other files it gives its agents.
- AGENTS.mdAGENTS.md
- docs/agents.mdAGENTS.md
- .cursor/rules/vibe-learn-craft.mdcCursor rule
- .cursor/rules/xrk-core-code.mdcCursor rule
- .cursor/rules/xrk-dev-requirements.mdcCursor rule
- .cursor/rules/xrk-infrastructure.mdcCursor rule
- .cursor/rules/xrk-project.mdcCursor rule
- .cursor/rules/xrk-third-party-plugins.mdcCursor rule
- accessibility-complianceSkill
- design-system-patternsSkill
- fronted-designSkill
- immersive-short-videoSkill
- interaction-designSkill
- mobile-android-designSkill
- mobile-ios-designSkill
- react-native-designSkill
- responsive-designSkill
- ui-ux-pro-maxSkill
- visual-design-foundationsSkill
- web-component-designSkill
- xrk-agent-runtimeSkill
- xrk-app-devSkill
- xrk-authSkill
- xrk-coding-styleSkill
- xrk-configSkill
- xrk-crawlSkill
- xrk-dockerSkill
- xrk-docsSkill
- xrk-github-researchSkill
- xrk-http-apiSkill
- xrk-infrastructureSkill
- xrk-llmSkill
- xrk-mcpSkill
- xrk-node-runtimeSkill
- xrk-pluginsSkill
- xrk-project-overviewSkill
- xrk-rendererSkill
- xrk-runtime-utilSkill
- xrk-subserverSkill
- xrk-system-coreSkill
- xrk-taskerSkill
- xrk-v3-apiSkill
- xrk-www-compatSkill
The file
Files
读者:Coding Agent(改
src/core)。产品模型工具用法见.xrk/skills/core/agent-tools。
文档与代码
- 契约:docs/agent-context.md · docs/ai-workflow.md · docs/harness-module-loop.md
- 代码:
src/infrastructure/ai-workflow/ai-workflow.js·harness-module-loop.js·loader.js·chat-pipeline.js
工作流
- 路径:
core/*/workflow/*.js(AiWorkflowLoader扫描,不用ai-workflow.streamDir) - 配置:
data/server_bots/{port}/ai-workflow.yaml;schemacommonconfig/system/system-ai-workflow.js - 工具:
registerMCPTool→this.mcpTools;远程挂载:export function getMcpServers()或 yamlmcp.remote - Tool 环:
callAI//v1+workflows →@xrkseek/harness(runHarnessModuleLoop) - 工厂:仅单次
chat/chatStream(无 MCP 多轮) - 出站:
prepareOutboundMessages= contextWindow trim - 策略/安全:
policies[]·security.toolScan·security.approval;执行门禁在MCPServer.handleToolCall - 斜杠:
slash-commands.js(/recipe);microagents:trigger-microagents.js - Shell:
#utils/exec-async.js;禁止promisify(exec)(skillxrk-node-runtime)
与子服务端关系
子服务端为可选 Python 扩展;AiWorkflow 核心链路不依赖子服务。业务 Core 可通过 AgentRuntime.callSubserver 调用子服务 apis/ 接口。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

