Agentforce · Agent Builder

Test Agentforce Agents

Check that an Agentforce agent behaves as intended by running it against test scenarios, including Agentforce Testing Center suites and scripted demo conversations, and reporting what passed and what did not.

107 tasks
Method

How MeshMesh does it

  1. Retrieves the agent definition and reads its topics and context variables.
  2. Inspects the data the agent relies on, finds or creates sample records to test with.
  3. Writes test cases as a YAML spec and deploys it to the org as an AiEvaluationDefinition for Testing Center.
  4. Starts test runs and writes the results up as an HTML report.
  5. Where the user runs the tests, drafts test prompts across the agent's topics plus a multi-turn scenario, tied to seeded sample records.

When a test turns up a fault, the next step is usually Diagnose Agentforce Agent Errors.

Products

Products and features it works with

Bars show how many of this capability's tasks touched each product.

  1. 1Agentforce AnalyticsAgentforce3 tasks