12

Prompt examples

Example prompts for using Umtri together with your AI agent.

Check your project grounds

“Check my Umtri projects.”

Calls list_projects to check your registered project grounds and replies with a list of slugs and last activity times. Before you start developing, you can find the ground that matches your project.

Inspect a tree and its open bugs

“Check the myProject ground and tell me what bugs are still open.”

Calls get_graph(slug: "myProject") and list_bugs(slug: "myProject", status: "active"), then reports a summary of how the project has grown and its active bugs.

File a bug (Write token required)

“Register a new bug on myProject: points aren’t restored after a booking is refunded. Review the risk and which node it belongs to with me.”

Works out the node, the impact and the risk from the existing tree and codebase, then calls create_bug like this:

{
  "slug": "myProject",
  "title": "Points aren't restored after a booking is refunded",
  "target": { "kind": "ground" },
  "score": 6
}

If your client says it lacks a tool or returns a 403, check the token scope. read tokens cannot call any write tool.

Hand over a growth plan with Plan Mode

“Check what’s drawn as plan nodes on the payments limb. Tell me if any part of the plan needs changing.”

  1. Calls get_graph and picks out the plan: true nodes
  2. Reviews feasibility against the existing code and tree structure
  3. Reports proposed changes: asks the user to confirm through plan nodes, plan edges and so on
  4. After the user’s review, implements them and confirms the growth (commit)

Transplant an existing project

“Analyze the current codebase and transplant it into the newly created myNewProject ground.”

  1. Calls get_graph(slug: "myNewProject") to check the transplanting status
  2. Transplants based on the existing codebase, discussing it with the user
  3. Once everything has moved over, asks the user to Root the ground