Free AI-201 – Salesforce Certified Agentforce Specialist Practice Test Question

Select an option, then click Submit answer.

• ○

  Build a Data Cloud search index and retriever over the opportunity data

• ○

  Add the record fields as merge field resources directly in the prompt template

• ○

  Write an Apex class that queries the records and returns a formatted string

• ○

  Create a Flow that calls an external API to fetch the account details

Select an option, then click Submit answer.

• ○

  Zero-data retention agreements with the LLM provider

• ○

  Data masking of personally identifiable information

• ○

  The audit trail stored in Data Cloud

• ○

  Prompt defense, which applies system-level policies that help the model resist prompt injection and jailbreak attempts

Select an option, then click Submit answer.

• ○

  The Data Library exposes full manual control over embedding models and chunk boundaries

• ○

  The Data Library provides a simplified, managed way to add sources because the platform automatically creates and maintains the underlying index and retriever

• ○

  The Data Library bypasses Data 360 entirely and stores content in the transactional database

• ○

  The Data Library allows agents to answer without any grounding content being indexed

Select an option, then click Submit answer.

• ○

  To store chunks together with their vector embeddings so content can be found by semantic similarity at query time

• ○

  To act as the system of record for transactional customer data

• ○

  To replace data streams as the ingestion mechanism for external sources

• ○

  To generate the agent's responses in place of the language model

Select an option, then click Submit answer.

• ○

  Profiles, permission sets, page layouts, and validation rules

• ○

  Triggers, classes, batch jobs, and scheduled jobs

• ○

  A config and system definition, variables, topics with instructions, and actions the agent can run

• ○

  Reports, dashboards, list views, and Kanban boards

Select an option, then click Submit answer.

• ○

  Add a note to the topic instructions asking the agent to verify identity first.

• ○

  Train the underlying model on examples of verified conversations.

• ○

  Remove the refund topic and handle refunds only through email.

• ○

  Apply a filter to the refund action so it is only available when the identity-verified variable is true.

Select an option, then click Submit answer.

• ○

  Capture the action's output into a variable, then reference that variable in the agent's message and in the follow-up action's input.

• ○

  Let the LLM remember the number from the conversation text and retype it.

• ○

  Create a new topic for case confirmations.

• ○

  Store the number in the classification description of the case topic.

Select an option, then click Submit answer.

• ○

  Apex actions are too slow for conversational use.

• ○

  The action should be rebuilt as a prompt template instead.

• ○

  Apex code is not supported for agent actions.

• ○

  Custom code can execute outside the protections of the caller's sharing and permissions, so the action must explicitly enforce access checks appropriate to the agent's security context before returning data.

Select an option, then click Submit answer.

• ○

  Have testers manually repeat the conversations in Agent Builder each night.

• ○

  Drive the conversations through Agent API from the test harness, since it supports programmatic sessions and message exchange.

• ○

  Export the agent script and diff it nightly.

• ○

  Use the email channel and parse the replies.

Select an option, then click Submit answer.

• ○

  Whether the change set accidentally deleted the template from production.

• ○

  Whether production users are still on legacy Salesforce licenses that predate prompt templates.

• ○

  Whether the deployed template version has been activated in production, because deployed prompt templates arrive in an inactive (draft) state.

• ○

  Whether the sandbox template needs to be deleted and recreated so it can be deployed again.

Select an option, then click Submit answer.

• ○

  Total number of topics configured on the agent

• ○

  Average number of actions available per topic

• ○

  Total Agentforce conversations, including abandoned sessions

• ○

  Deflection rate — the share of conversations the agent resolves end-to-end without escalation to a human agent

Select an option, then click Submit answer.

• ○

  MCP and A2A are competing standards, and an architecture must choose one or the other

• ○

  MCP lets agents reach external tools and data sources, while A2A lets agents delegate tasks and collaborate with other agents; the two are complementary

• ○

  A2A connects agents to databases and APIs, while MCP connects agents to each other

• ○

  Both protocols are Salesforce-proprietary standards used only inside the Salesforce platform