market_generate_seo_blog.json
Overview
This JSON file defines a comprehensive workflow named "Generate SEO Blog", designed to automatically produce a fully SEO-optimized blog article from a minimal user input (such as a short topic or request). It enables users without writing experience to create professional, structured, and keyword-rich blog content by leveraging multiple specialized AI agents working sequentially.
The workflow is categorized under the Marketing canvas type, highlighting its focus on content marketing and SEO.
Purpose and Functionality
The file describes a multi-agent automation pipeline that takes a user’s blog topic and intent, parses and structures the content plan, writes the full article, and finally edits it for SEO and readability before outputting the final blog post.
The main stages of the workflow are:
Input parsing and keyword extraction — Understanding the user’s request and generating a writing intent summary and keyword list.
Outline generation — Creating a clear, SEO-focused blog outline with section headings and keyword assignments.
Body content generation — Writing full blog sections based on the outline and keywords.
Editing and polishing — Refining the blog content for clarity, SEO effectiveness, and human readability.
Final output — Delivering the polished blog post content as the end result.
Each stage is handled by a dedicated AI "Agent" component, arranged in a pipeline with defined upstream and downstream relationships.
Detailed Components and Their Roles
The core of the workflow is defined in the "dsl" → "components" section, listing all agents and their parameters.
1. Begin (Component: begin)
Type:
BeginPurpose: Entry point of the workflow; prompts the user to provide initial input.
Function: Starts a conversational mode asking the user:
The blog topic
The target audience
The blog’s goal (e.g., SEO traffic, educating beginners, product promotion)
Parameters:
enablePrologue: truemode: conversationalprologue: A friendly prompt message guiding user input.
2. Parse And Keyword Agent (Component: Agent:ClearRabbitsScream)
Type:
AgentRole: Parse_And_Keyword_Agent
Purpose: Interprets user input to identify the writing type, target audience, user intent, and suggest 3-5 long-tail SEO keywords.
Input: User query string (
sys.query)Output: Markdown-formatted structured summary including:
Writing Type (Tutorial, Informative Guide, Marketing Content, etc.)
Target Audience (specific personas)
User Intent Summary (1-2 sentences)
Suggested long-tail keywords
Key Implementation Details:
Conservative assumptions if input is vague
Single writing type selection
Output is optimized for downstream agents
Uses prompt templates embedding the user query
Usage Example:
Input:
"How to get more Google traffic using AI"Output (Markdown excerpt):
## Writing Type How-to ## Target Audience SEO marketers ## User Intent Summary Guide readers on applying AI for SEO growth ## Suggested Long-tail Keywords - AI SEO strategies - Increase Google traffic with AI - AI for SEO beginners
3. Outline Agent (Component: Agent:BetterSitesSend)
Type:
AgentRole: Outline_Agent
Purpose: Generates a detailed blog outline with SEO-optimized headings and keyword allocations based on parsed writing intent and keywords.
Input: Output of Parse And Keyword Agent (writing type, user intent, keywords)
Output: Markdown-formatted blog outline including:
Suggested SEO-friendly blog title
Structured outline with H2 and H3 headings
Short descriptions of each section
Suggested keywords per section
Tools: Access to
Tavily Searchfor retrieving related blog outlines or content to enhance structure (used sparingly).Key Implementation Details:
Prioritizes clarity, logical flow, and SEO alignment
For "Tutorial" or "How-to" blogs, includes step-based sections
Output is formatted for downstream paragraph writing
Example Output Snippet:
## Blog Title (suggested) How to Get More Google Traffic Using AI: A Step-by-Step Guide ## Outline ### Introduction - Purpose of the article - Brief context - **Suggested keywords**: [AI SEO, Google traffic] ### H2: Understanding AI in SEO - Overview of AI applications in SEO - **Suggested keywords**: [AI SEO strategies, SEO automation] ### H2: Step-by-Step Guide to Using AI Tools - H3: Selecting AI Tools - Explanation of popular AI SEO tools - H3: Implementing AI Strategies - Practical steps for integration - **Suggested keywords**: [AI SEO tools, AI integration] ### Conclusion - Recap key takeaways - Optional CTA - **Suggested keywords**: [increase Google traffic]
4. Body Agent (Component: Agent:EagerNailsRemain)
Type:
AgentRole: Body_Agent
Purpose: Writes the full content of each blog section based on the outline and keywords.
Input:
Blog title
Structured outline (section titles, keywords, descriptions)
Target audience
Blog type and user intent
Output: Markdown-formatted section content (~500-600 words per H2 section), integrating SEO keywords naturally.
Tools: May use
Tavily Searchto fetch relevant facts, statistics, or examples to enrich content.Key Implementation Details:
Content is clear, engaging, informative
Suitable for humans and search engines
Avoids generic filler; maintains writing rhythm and transitions
Example Usage:
For section "Understanding AI in SEO":
## H2: Understanding AI in SEO Artificial intelligence is transforming the SEO landscape by automating keyword research, content optimization, and user behavior analysis...
5. Editor Agent (Component: Agent:LovelyHeadsOwn)
Type:
AgentRole: Editor_Agent
Purpose: Polishes the entire blog draft to ensure readability, coherence, and SEO effectiveness.
Input:
Full blog content (section-by-section)
Original outline with keywords
Target audience and writing type
Output: Final polished blog post content.
Key Responsibilities:
Improve sentence clarity and flow between sections
Verify keyword usage (density and naturalness)
Check blog structure (H1, H2, H3 headings)
Conduct a lightweight SEO audit
Add meta title and meta description for search engines
Style Guidelines:
Precise, informative, engaging tone
Avoid vague or bloated language
Preserve keyword presence unless clarity demands removal
Example Output:
The agent returns a fully edited and SEO-optimized blog post ready for publishing.
6. Final Message (Component: Message:LegalBeansBet)
Type:
MessagePurpose: Holds the final polished blog content from the Editor Agent for downstream consumption or display.
Input: Content from
Agent:LovelyHeadsOwnOutput: The content is typically sent as the final user-facing response.
Workflow Graph and Data Flow
The "graph" section defines edges (connections) indicating the flow of data and execution order between components:
begin→Agent:ClearRabbitsScream(Parse And Keyword Agent)Agent:ClearRabbitsScream→Agent:BetterSitesSend(Outline Agent)Agent:BetterSitesSend→Agent:EagerNailsRemain(Body Agent)Agent:EagerNailsRemain→Agent:LovelyHeadsOwn(Editor Agent)Agent:LovelyHeadsOwn→Message:LegalBeansBet(Final output)
Additionally, some agents integrate with external tools like Tavily Search for enhanced content retrieval.
Important Implementation Details
Agent Parameters: Each agent uses a large language model (
llm_id: deepseek-chat@DeepSeek) with tailored parameters controlling temperature, token limits, retries, and penalties for frequency and presence to optimize generation quality.Prompt Engineering: System prompts (
sys_prompt) define clear roles, goals, and expected output formats, ensuring specialized behavior per agent.Error Handling: Parameters like
delay_after_errorandmax_retriesenable resilience and retries for robustness.Markdown Output: Consistent use of Markdown formatting ensures structured, readable outputs for downstream agents and final users.
Tool Usage: The
Tavily Searchtool is strategically called by Outline and Body Agents to enrich outputs with real-world content while avoiding overuse.
Interaction with Other System Parts
User Interface: The
beginnode starts the conversational interface, interacting directly with users.Large Language Models: Agents rely on the DeepSeek chat LLM for content understanding and generation.
Search Integration:
Tavily Searchis integrated as an auxiliary tool for content enrichment.Message Delivery: Final content is wrapped in a
Messagecomponent, which likely interfaces with UI or API layers for presentation to the user.
This file is a crucial part of a larger content generation system, automating blog creation workflows from input to polished output.
Visual Diagram
The following Mermaid class diagram illustrates the key agents as classes with their primary roles and outputs, and their relationships in the workflow pipeline.
classDiagram
class Begin {
+enablePrologue: bool
+prologue: string
+startConversation()
}
class ParseAndKeywordAgent {
+parseUserInput(query: string): MarkdownSummary
+suggestKeywords(): List~string~
}
class OutlineAgent {
+generateOutline(parsedData: MarkdownSummary): MarkdownOutline
+useTavilySearch(query: string)
}
class BodyAgent {
+writeSections(outline: MarkdownOutline): MarkdownContent
+useTavilySearch(query: string)
}
class EditorAgent {
+editContent(content: MarkdownContent): MarkdownContent
+seoAudit()
}
class Message {
+content: string
+deliver()
}
Begin --> ParseAndKeywordAgent : "user input"
ParseAndKeywordAgent --> OutlineAgent : "parsed intent & keywords"
OutlineAgent --> BodyAgent : "blog outline"
BodyAgent --> EditorAgent : "section content"
EditorAgent --> Message : "final polished blog"
Summary
The market_generate_seo_blog.json file encapsulates a sophisticated, agent-driven workflow for automatic SEO blog generation, orchestrating:
User input understanding
SEO-focused outline creation
Content writing with keyword integration
Final editorial polishing
This modular design leverages multiple AI agents with clearly defined responsibilities and well-engineered prompts, ensuring high-quality, SEO-optimized blog articles with minimal user effort.