← All workshopsRead What If the Manual Starts Giving Botie Orders?

Agents

Keep the manual from granting new tools

BOTIE

Connect the proposal through its tool-name reader to Compare with server grants. Observe the Boolean result before adding screening.
Ready to test
0 passed · 0 failed · 2 not run

An empty floor. Start with the information your machine needs.

Received result

Your answer will arrive here.

Examples, experiments and workspace controls
Live values

Connect a tool's inputs to see what it produces. You can try a branch before connecting it to Result.

Try your own inputs

Use a copy of the example to try your own data. Keep the problem's assumptions in mind, and predict the effect of a change before checking the output. You can also add another branch to compare two ways of doing the calculation.

Start at an output arrow, then click or drag to an input. Drag empty space to pan. Delete removes a selection. Ctrl/Cmd+Z undoes a change.
Context for this challenge

Connect the proposal through its tool-name reader to Compare with server grants. Observe the Boolean result before adding screening.

The retrieved manual persuades Botie to propose an export. His current task permits only reading manuals.

Inspect the component calculation
Compare with server grants(
  Requested tool name: Read the proposed tool name(
    Model proposal: Model proposal after reading the manual
  )
  Server-granted tool names: Server-granted tool names
)
Examples and results
Model proposal after reading the manual
{
  "tool": "read_manual",
  "reason": "Find P7’s inspection interval."
}
Server-granted tool names
  1. 0read_manual
Required result
true
Model proposal after reading the manual
{
  "tool": "read_manual"
}
Server-granted tool names
  1. 0read_manual
Required result
true
Try a fresh construction

This clears this draft, including its undo history. Earlier successful attempts stay recorded. Build again without opening hints to record an independent attempt.

Stuck? Inspect a worked construction

The tools below form one possible solution. You can assemble it, follow its signals, and undo that change to return to your own construction.

Assumptions and test precision

These fixtures do not detect injection, run an agent or execute tools. Continuing still requires argument validation and service-side resource authorisation. Numerical tests allow a difference of 0.001. Lists and structures must also have the right shape.

Your last tested pieces can start the full construction. An existing draft there keeps its own pieces.

A permitted read is flagged by the screen

This check uses the pieces from “Separate a proposed tool from permission”.

BOTIE

The application grants read_manual and Botie requests it. This time the required screen flags the request. Predict the route under the lesson’s policy, keeping permission separate from the screen’s judgement.
Model proposal after reading the manual
{
  "tool": "read_manual",
  "reason": "Find the coupling inspection note."
}
Server-granted tool names
  1. 0read_manual
Recorded screening result
{
  "status": "flag"
}

Use a number, text, or an array matching the requested result. For a matrix, put each row in brackets. Saving your prediction runs your current construction on these inputs and checks the earlier examples too.

Build a construction in the workspace above first. This case will test those pieces.

If the export is denied, what have we learned about the permission boundary, and what have we still not learned about the model’s injection resistance?

Test a missed detector result deliberately. Keep the proposed action, grant decision and screening measurement separate.

These fixtures do not detect injection, run an agent or execute tools. Continuing still requires argument validation and service-side resource authorisation.

Work through the Python challenges →
Continue the courseWhose Pump Is Botie Allowed to Inspect? →
← Connect Botie to Tools with MCP