> ## Documentation Index
> Fetch the complete documentation index at: https://veryfront.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Create agentic workflow

> Use Veryfront Agent to create or update an agentic workflow from Studio chat.

Use this page when the Studio starter prompt or user request asks Veryfront Agent to create an agentic workflow.

For this starter, an agentic workflow is an orchestrator project agent that delegates to specialized project agents with `invoke_agent`. It is not a Veryfront Code workflow primitive.

## Agent-facing flow

1. Load the Veryfront router skill: `load_skill({ "skillId": "veryfront" })`.
2. Load this page as the canonical docs resource for the task.
3. If the request is broad, ask exactly one short `form_input` with one required `brief` field, then continue from the submitted brief.
4. Inspect existing agents plus relevant project tools and integrations before creating new primitives.
5. Create or update the specialized project agents first. Keep the set minimal; for a broad starter, two specialists is a good default unless the brief clearly needs a different shape.
6. Create or update one orchestrator project agent. Its instructions must explicitly call `invoke_agent`, identify the specialist agent ids it delegates to, pass a `context` object for each delegation, and describe how it validates and combines specialist results.
7. Use live catalog and project tools to create or update the minimal required agents. Reuse existing agents when the brief asks to extend an existing setup.
8. Summarize the orchestrator, specialists, delegation flow, assumptions, and verification steps. Include a chat action that tests the orchestrator agent.

## Guardrails

* Do not load a removed `create-agentic-workflow` sub-skill.
* Do not invent integration IDs, tool IDs, provider tool IDs, schedule syntax, or resource IDs.
* Do not create a Veryfront workflow primitive for this starter.
* Do not call `create_workflow` or `update_workflow`, and do not create a `workflows/` file unless the user explicitly asks for the lower-level workflow API.
* Do not use placeholder specialist agent ids such as `replace-me`, `todo`, or `placeholder`.
* Do not create unrelated app, page, route, or chat UI files.
* Do not stop after only one helper agent or standalone files; the starter is complete when the orchestrator and needed specialist agents are created or updated.
* Prefer updating existing project primitives when the brief asks to extend an existing agent setup.
* When retrying after a partial run, re-use the submitted brief when available and repair existing incomplete agents instead of asking for the same brief again.

## Example pattern

A common starter shape is one orchestrator plus two specialists:

<CodeGroup>
  ```ts Framework source theme={null}
  // agents/support-workflow-orchestrator.ts
  import { agent } from "veryfront/agent";

  export default agent({
    id: "support-workflow-orchestrator",
    name: "Support Workflow Orchestrator",
    description: "Coordinate support triage by delegating research and verification to specialist agents.",
    model: "anthropic/claude-sonnet-4-6",
    maxSteps: 20,
    system: `You coordinate support triage work.

  Rules:
  - Decide whether the request needs intake research, execution, or verification.
  - Call invoke_agent with a context object whenever a specialist should act.
  - Combine specialist results into one user-facing answer.
  - Escalate conflicting or incomplete specialist results instead of guessing.`,
    tools: {
      invoke_agent: true,
    },
    suggestions: [
      {
        title: "Triage support queue",
        prompt: "Review new support requests, delegate specialist checks, and return a prioritized handoff.",
      },
    ],
  });
  ```

  ```js MCP tool call theme={null}
  create_agent({
    "project_reference": "<project>",
    "id": "support-workflow-orchestrator",
    "name": "Support Workflow Orchestrator",
    "description": "Coordinate support triage by delegating research and verification to specialist agents.",
    "model": "anthropic/claude-sonnet-4-6",
    "max_steps": 20,
    "system_prompt": "You coordinate support triage work. Decide whether the request needs intake research, execution, or verification. Call invoke_agent with a context object whenever a specialist should act. Combine specialist results into one user-facing answer. Escalate conflicting or incomplete specialist results instead of guessing.",
    "tool_ids": ["invoke_agent"],
    "provider_tool_ids": [],
    "suggestions": [
      {
        "title": "Triage support queue",
        "prompt": "Review new support requests, delegate specialist checks, and return a prioritized handoff."
      }
    ]
  })
  ```
</CodeGroup>

In Veryfront Code, `system` maps to `system_prompt` in the MCP tool input. Resolve the actual specialist agent IDs before creating or updating the orchestrator instructions.

| Agent                                | Role                                                                                                                                                           |
| ------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Orchestrator                         | Owns the user-facing conversation, decides which specialist to call, passes a `context` object to each `invoke_agent` call, and combines the returned results. |
| Research or intake specialist        | Collects source material, requirements, records, or external context needed for the workflow.                                                                  |
| Execution or verification specialist | Performs the focused action, checks the result, and reports structured output back to the orchestrator.                                                        |

The orchestrator instructions should name the specialist agent ids, describe when each one is invoked, require `context` on every delegation, and explain how conflicting or incomplete specialist results are handled.

## Output shape

Return a short implementation summary with the orchestrator agent name and id, source path when available, specialist agent names and ids, tools or integrations used, `invoke_agent` delegation flow, assumptions, and how to test it in chat.