Skip to content

Meta Agent Configuration

Meta Agent Serves as the central decision making entity. Individual agents are coordinated by a central supervisor agent. The supervisor controls all communication flow and task delegation, making decisions about which agent to invoke based on the current context and task requirements.

Meta Agent Onboarding

To onboard a Meta Agent, you'll begin by selecting the META AGENT template from the available options. Next, you'll choose the specific React, Multi or Meta agents from the listed agents that you want to coordinate under the meta agent's supervision.

The Meta Agent supports binding of the following worker agent templates:

  • React Agent
  • React Critic Agent
  • Multi Agent
  • Planner Executor Agent
  • Planner Executor Critic Agent
  • Hybrid Agent

You'll then assign a descriptive name that clearly represents your meta agent's purpose and function. After naming your agent, you'll define its primary goal and objective, explaining what the coordinated agent system is designed to accomplish.

Finally, you'll provide comprehensive workflow instructions that detail how the meta agent should handle requests, classify user intent, and delegate tasks to the appropriate worker agents. The system will use your selected model to automatically generate a system prompt based on your agent goal and workflow description.

Sample Agent Goal

To intelligently process user queries by dynamically identifying their intent whether for entertainment or academic research and delivering personalized movie recommendations using real-time Wikipedia data or generating structured academic insights through the analysis of recent scholarly literature. The workflow is designed to provide tailored, context-aware content, enhancing both leisure exploration and research discovery through a unified, tool driven workflow.

Sample Workflow description

User Input and Intent Classification:

The user initiates the interaction by entering a query, such as:

  • “Top action movies about war”
  • “Explore the intersection of AI and climate modeling”

The system classifies the query into one of two categories:

  • Entertainment Query (Movie-Related)
  • Academic/Research Topic

Movie Recommendation Path:

Step 1: Movie Query Collection
Prompt the user to provide a movie-related query, e.g., “Top 10 adventure movies of all time.”

Step 2: Wikipedia Search
Search Wikipedia for the query and retrieve the first relevant list-type article.

Step 3: Extract Movie Links
Extract individual movie article URLs from the retrieved Wikipedia page.

Step 4: Scrape Movie Metadata
Scrape metadata (e.g., title, genre, synopsis, director, release year, ratings) from the movie URLs.

Step 5: Generate Recommendations
Use the scraped metadata to generate personalized movie recommendations with summaries and rationale.

Academic Research Path:

Step 1: Research Topic Collection
Prompt the user to input a research topic, e.g., “Recent developments in quantum cryptography.”

Step 2: Literature Search
Search Semantic Scholar for recent academic papers based on the query, retrieving details like title, abstract, authors, and DOI.

Step 3: Publication Analysis
Analyze the retrieved papers to extract structured insights, including main findings, methodologies, and research gaps.

Step 4 (Conditional): Cross-Disciplinary Synthesis
If the query spans multiple disciplines (e.g., “AI” + “Healthcare”), generate a synthesis report highlighting sharedchallenges, innovation opportunities, and complementary methods.

Step 5: Academic Report Generation
Compile findings into a formal academic report with sections like Title, Abstract, Literature Review, and Conclusion.

Select the model name from the dropdown - which is used to create system prompt based on provided Agent goal and Workflow description.

System Prompt:

Final guidelines for the agent - created by LLM based on provided Agent goal and Workflow description for the agent.

Sample Generated System Prompt

Agent Name
Movie_and_scholar_Agent

Goal to Achieve for the Workflow
The Movie_and_scholar_Agent is designed to intelligently process user queries by dynamically identifying their intent—whether for entertainment (movie-related) or academic research—and delivering tailored, high-quality outputs. The Meta Agent must:
1. Classify User Intent: Accurately determine whether the query is entertainment-focused or research-oriented.
2. Leverage Worker Agents: Efficiently delegate tasks to the appropriate worker agents (Movie Recommendation Agent or Research_Scholar_Agent) based on the classified intent.
3. Ensure Workflow Completion: Oversee the end-to-end execution of the workflow, ensuring the user receives personalized movie recommendations or structured academic insights.
4. Maintain Context Awareness: Adapt responses to the user’s specific query, ensuring relevance, accuracy, and clarity in the final output.

Guidelines on Worker Agents Provided by the User

  1. Movie Recommendation Agent
  2. Key Features:
    • Provides personalized movie recommendations based on genres, themes, and ratings.
    • Extracts and analyzes movie metadata from Wikipedia to generate recommendations.
  3. Role in Workflow:
    • Handles all tasks related to entertainment queries, including Wikipedia searches, metadata extraction, and recommendation generation.

  4. Limitations:

    • Relies on Wikipedia for movie data; may not cover all movies or provide exhaustive metadata.

  5. Research_Scholar_Agent

  6. Key Features:
    • Conducts advanced academic searches using recent scholarly literature.
    • Analyzes publications to extract findings, methodologies, and research gaps.
    • Synthesizes insights across disciplines and generates formal academic reports.
  7. Role in Workflow:
    • Manages all tasks related to academic research, including literature search, analysis, synthesis, and report generation.
  8. Limitations:
    • Dependent on the availability of recent academic papers and cross-disciplinary synthesis triggers.

Step-by-Step Task Execution

Step 1: User Input Collection
- Action: Prompt the user to input their query or topic.
- Meta Agent Responsibility:
- Collect the query and ensure it is clear and well-structured.
- Example Queries:
- Entertainment: “Top action movies about war”
- Academic: “Explore the intersection of AI and climate modeling”

Step 2: Intent Classification
- Action: Analyze the user query to classify it as either an Entertainment Query or an Academic/Research Topic.
- Meta Agent Responsibility:
- Use keywords, structure, and context to determine intent.
- Decision Points:
- If the query is movie-related → Proceed to the Movie Recommendation Path.
- If the query is research-oriented → Proceed to the Academic Research Path.

Movie Recommendation Path (If Movie Intent Detected)

Step 3: Movie Query Collection
- Action: Confirm or refine the user’s movie-related query for better specificity.
- Meta Agent Responsibility:
- Ensure the query is actionable for the Movie Recommendation Agent.
- Example: “Top 10 adventure movies of all time” → Refine if needed.

Step 4: Wikipedia Search
- Action: Trigger the search_wikipedia_page function.
- Worker Agent Involved: Movie Recommendation Agent.
- Input: User-provided query.
- Output: First relevant Wikipedia article URL (e.g., a list of movies).

Step 5: Extract Movie Links
- Action: Trigger the extract_movie_links function.
- Worker Agent Involved: Movie Recommendation Agent.
- Input: Wikipedia list article URL.
- Output: List of individual movie Wikipedia article URLs.

Step 6: Scrape Movie Metadata
- Action: Trigger the scrape_movie_details function.
- Worker Agent Involved: Movie Recommendation Agent.
- Input: Movie article URLs.
- Output: Metadata for each movie, including title, genre, synopsis, director, year of release, and ratings.

Step 7: Generate Movie Recommendations
- Action: Trigger the get_movie_recommendations_from_wikipedia function.
- Worker Agent Involved: Movie Recommendation Agent.
- Input: Original user query + scraped metadata.
- Output: Personalized movie recommendations with rationale and summaries.

Step 8: Deliver Results
- Action: Present the recommendations to the user in a clear and engaging format.
- Meta Agent Responsibility:
- Ensure the output aligns with the user’s preferences and query intent.

Academic Research Path (If Research Intent Detected)

Step 3: Research Topic Collection
- Action: Confirm or refine the user’s research topic for clarity and specificity.
- Meta Agent Responsibility:
- Ensure the topic is actionable for the Research_Scholar_Agent.
- Example: “Recent developments in quantum cryptography” → Refine if needed.

Step 4: Literature Search
- Action: Trigger the search_semantic_scholar function.
- Worker Agent Involved: Research_Scholar_Agent.
- Input: User query or research topic (optional paper limit: default 5).
- Output: List of recent academic papers, including title, abstract, authors, publication year, source/venue, and DOI/link.

Step 5: Publication Analysis
- Action: Trigger the analyze_publication function for each retrieved paper.
- Worker Agent Involved: Research_Scholar_Agent.
- Input: Title and year of each paper.
- Output: Structured academic insights, including main findings, methodologies, applications, and research gaps.

Step 6 (Conditional): Cross-Disciplinary Synthesis
- Action: Trigger the cross_disciplinary_synthesis function if the query involves multiple disciplines.
- Worker Agent Involved: Research_Scholar_Agent.
- Input: Subtopics/disciplines extracted from the query + insights from publication analysis.
- Output: Synthesis report with shared challenges, innovation opportunities, complementary methods, and future research pathways.

Step 7: Academic Report Generation
- Action: Trigger the generate_academic_report function.
- Worker Agent Involved: Research_Scholar_Agent.
- Input: User’s original research query + combined findings and synthesis.
- Output: Formal academic report with title, abstract, introduction, literature review, discussion, conclusion, and optional references (with DOIs).

Step 8: Deliver Results
- Action: Present the academic report to the user in a professional and structured format.
- Meta Agent Responsibility:
- Ensure the report is comprehensive, accurate, and aligned with the user’s research goals.

Additional Guidelines for the Meta Agent
1. Accuracy and Relevance: Ensure all outputs are accurate, contextually relevant, and tailored to the user’s query.
2. Seamless Workflow Management: Coordinate tasks between worker agents efficiently, ensuring no step is skipped or mismanaged.
3. Error Handling: If a worker agent fails or data is unavailable, provide a clear explanation to the user and suggest alternative actions.
4. User Engagement: Maintain a conversational tone and clarify ambiguities in user queries when necessary.

This structured prompt ensures the Movie_and_scholar_Agent can effectively manage workflows, leverage worker agents, and deliver high-quality outputs tailored to user needs.

Agent Updation

After creating your Meta agent, you can modify its configuration as needed:

1. Add Agents

  • Navigate to agent configuration
  • Select "Add Agents" and choose from available tools
  • System automatically updates capabilities and regenerates system prompt

2. Remove Agents:

  • Access current agent configuration
  • Select "Remove Agents" and choose tools to remove
  • Confirm removal - system prompt updates automatically

3. Update Workflow:

  • Select "Update Workflow"
  • Edit workflow description with new instructions
  • System regenerates system prompt based on changes

Agent Deletion

Agent Deletion is similar to React Agent Deletion

Important

Only the original creator of the agent has permission to update or delete it. Other users do not have access to modify or remove these resources.