agenticflow-skills

What This Skill Does

The AgenticFlow skill serves as the foundational toolkit for the OpenClaw AI agent to design, configure, and execute sophisticated automation pipelines. It enables users to move beyond simple chat interactions by constructing complex architectures including linear workflows, autonomous intelligent agents with defined personas, and multi-agent workforce systems. Whether you need to integrate external APIs via MCP providers, process data through specialized nodes, or orchestrate high-level collaborative patterns like Supervisor or Swarm models, this skill provides the schema-driven framework to manage the entire lifecycle of your automation logic.

Installation

To integrate this capability into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/seanphan/agenticflow-skill Once installed, the agent will have access to all documentation, node type schemas, and orchestration patterns defined within the AgenticFlow ecosystem, allowing for immediate architectural planning.

Use Cases

• Automated Data Pipelines: Create a linear workflow that triggers on a webhook, scrapes a website, parses the JSON, and sends a summary to Slack.
• Agentic Specialization: Configure a 'Researcher' agent to search the web and a 'Writer' agent to draft content, orchestrated by a 'Supervisor' agent to ensure quality.
• Content Systems: Combine image generation nodes with text transformation nodes to automatically create and distribute branded marketing collateral.
• API Integration: Use the integrated MCP connectors to pull data from Notion or Gmail and perform conditional logic using workflow nodes.

Example Prompts

1. 'Build a linear workflow that fetches new entries from a Notion database, summarizes the content using an LLM node, and emails the result via the Gmail integration.'
2. 'Set up a Supervisor workforce pattern where one agent coordinates between a research agent and a proofreading agent to produce a final report.'
3. 'How do I configure a custom agent persona that uses the Claude Ask node as its primary reasoning engine?'

Tips & Limitations

• Linear Constraints: Remember that workflows in AgenticFlow are strictly linear and sequential. If you require branching logic or complex loops, you must design these as separate workforce orchestrations rather than single workflow nodes.
• Node Discovery: Always refer to the node-types.md reference to verify if a specific external service has a pre-built connector before attempting to build a custom http_request node.
• Orchestration: When using the Workforce pattern, ensure you have clearly defined the communication protocols between agents to prevent infinite loops in Swarm patterns.
• Error Handling: When building long chains, verify each node's output schema in overview.md to ensure data compatibility between sequential steps.