MakeUserStory for AI Agents
Connect your AI assistant directly to MakeUserStory via the Model Context Protocol (MCP). Generate user stories, search your project knowledge base, and export to GitHub or Jira — all without leaving your editor.
Why MCP changes everything
Agentic Workflows
Let your AI agent autonomously generate stories, decompose epics, and file issues as part of a multi-step workflow — no manual copy-pasting required.
Context-Aware Generation
MCP gives agents access to your project documents, so every generated story is grounded in your real domain knowledge and requirements.
Zero Context Switching
Works inside VS Code, Claude Desktop, Cursor, and any MCP-compatible client. Your ideas become backlog items without ever leaving your editor.
Get started in 2 steps
The MCP server uses Streamable HTTP transport — no local process to run, no stdio tunnelling. Add the URL, connect, and your client handles the rest.
Add the MCP server to your client
Open your AI client's MCP or connector settings and add the MakeUserStory server URL:
Each MCP client has its own way to add a server — check your client's documentation for the exact steps.
Authorize your account
When your MCP client first connects, the server automatically initiates an OAuth flow and opens the MakeUserStory login page in your browser. Sign in, approve access, and you're done — your client is authorized and ready to use all MCP tools.
No token copying or environment variables required. Authorization is handled entirely by the OAuth flow, and credentials are stored securely by your MCP client.
Start generating
Ask your AI assistant in natural language. For example:
"Generate a user story for adding dark mode support to the settings page"
"Search my project documents for authentication requirements, then generate a spike to evaluate OAuth providers"
"Decompose the checkout epic into individual user stories and export each one to GitHub"
"Refine the last generated story to include edge cases for expired sessions"
Available MCP tools
All tools are available to any MCP-compatible AI agent. Your agent will discover them automatically on connection.
| Tool | Description |
|---|---|
generate_issue | Generate a user story, bug report, spike, task, or epic from a natural language description. Returns a fully structured issue with title, description, acceptance criteria, and more. The response includes a thread_id that can be used with refine_issue for iterative improvements. |
decompose_issue | Break down a user story or epic into smaller, atomic work items. Returns a list of sub-stories or tasks. |
refine_issue | Refine an existing generated issue with a follow-up instruction. Requires the thread_id from a previous generate_issue call. Returns the updated issue content and the same thread_id for further refinement. |
get_thread | Retrieve the full conversation thread for a generated issue, including all refinement turns in order. Useful for reviewing the history of changes made to a story. |
search_documents | Search your uploaded project documents using semantic vector similarity. |
export_to_github | Create a GitHub issue directly from a generated story (requires GitHub connection). |
export_to_jira | Create a Jira issue from a generated story (requires Jira connection). |
list_github_repos | List the GitHub repositories connected to your account, so you can discover available export targets before running export_to_github. |
list_personas | Retrieve all available personas for your account, including their names, descriptions, and IDs. |
get_persona | Fetch a specific persona by its ID to view its full details. |
create_persona | Create a new persona with a name and description to tailor generated stories to a specific user perspective. |
edit_persona | Update an existing persona's name or description to keep it aligned with evolving project needs. |
Active MCP Clients
Loading clients…
MCP access requires an API-enabled plan
Any plan with API Access can connect to the MCP server. See the pricing page for details.
Ready to supercharge your workflow?
Add the MCP server URL to your AI client, sign in when prompted, and start letting your agent do the backlog grooming.