Starter prompts

Studio starter prompts start broad work from project chat. Veryfront Agent loads the veryfront router skill, reads the matching canonical docs page, then uses live project and catalog tools. Use the same source-first pattern across starter guides. Show the project source or saved artifact first, then show the related MCP tool call when a control-plane form matters.

Studio prompt	Canonical docs	Expected first actions
Create a plan	Create a plan	Load `veryfront`, read the docs, ask one `brief` field when details are missing
Make a research	Make a research	Load `veryfront`, read the docs, ask one `brief` field when details are missing
Create an agent	Create an agent	Load `veryfront`, read the docs, search integrations, create or update the agent
Create a skill	Create a skill	Load `veryfront`, read the docs, create or update a reusable skill

Runtime contract

Load load_skill({ "skillId": "veryfront" }) first.
Do not load removed sub-skills such as create-agent, plan, research, create-skill, or create-agentic-workflow.
Read only the canonical docs page needed for the active prompt.
Prefer live tool results for project state, integration IDs, tool IDs, files, agents, prompts, resources, and workflows.
Ask exactly one required freeform brief field for broad first-turn intake when details are missing.
Do not create app, page, route, or chat UI files unless the user asks for those artifacts.

Verification

Closing a starter prompt issue requires a Veryfront E2E test or staging run with one clean conversation per current prompt suggestion, visible output, no terminal error, and evidence linked from the issue.

Getting Started

Concepts

Guides

Runtime contract

Verification

​Runtime contract

​Verification

Runtime contract

Verification