Salesforce Certified Agentforce Specialist

Last checked October 22, 2026 against the official Salesforce exam page.

1. Question 1AI Agents

   A retailer wants an Agentforce agent on its website that answers order-status questions and can start a return for an order. The team wants the agent to use approved business steps rather than free-form decisions. Which configuration fits best?

   1. AGive the agent unrestricted access to every object and let it decide which records to change without any defined actions
   2. BAn agent with topics for order status and returns, plus Flow actions for each step
   3. CWrite a single prompt template that answers every question and turn off all actions so the agent only replies with text
   4. DBuild one Apex class that handles all customer chats in a single method with no topics or instructions for the agent
   Show answer and explanationHide answer

   Correct answer: B

   Topics define the scope of what the agent handles, and actions expose approved steps such as flows, which keeps behaviour bounded. Unrestricted access removes the control the team wants, a text-only prompt cannot perform the return, and a single Apex method with no topics gives no structure for the agent.

2. Question 2Development Lifecycle

   A team is moving an agent from a sandbox to production. The team wants the change to be repeatable, reviewed and reversible. Which approach fits best?

   1. ARecreate the agent by hand in production each time, copying settings from the sandbox screen by screen, with no record of the changes
   2. BEdit the agent directly in production, since the sandbox copy is always out of date and cannot be trusted at all
   3. CPackage the agent's configuration in source control and deploy through a reviewed pipeline
   4. DDeploy the agent once and never change it again, so that no change ever needs to be reviewed or reversed in the future
   Show answer and explanationHide answer

   Correct answer: C

   Packaging the configuration in source control and deploying through a reviewed pipeline with a rollback path makes the change repeatable and reversible. Manual recreation and direct production edits are error-prone and unrecorded, and never changing the agent prevents needed fixes.

3. Question 3Prompt Engineering

   A team builds a Prompt Builder template that drafts follow-up emails from case records. The drafts sometimes include a customer's full credit card number that was stored in a note. What should the team change first?

   1. AAsk the model to hide any numbers it sees by writing a note in the prompt that says not to show numbers to anyone
   2. BSend the drafts to a public blog so that customers can see the content and report any sensitive data they find
   3. CTurn the template into a free-text prompt that includes every field on the case record, so that the model sees everything
   4. DRemove or mask sensitive fields from the data passed to the prompt, and use the Einstein Trust Layer's data masking
   Show answer and explanationHide answer

   Correct answer: D

   The safest fix is to keep sensitive fields out of the prompt and use masking, so the model never receives them. A note asking the model to hide numbers is unreliable, passing every field increases exposure, and publishing drafts is a serious leak.

4. Question 4Data Cloud for Agentforce

   An agent must answer questions using customer purchase history, which lives in several source systems. Data Cloud should bring the data together for the agent. Which Data Cloud step creates a unified view of each customer?

   1. ACreating a dashboard that displays each source system in its own tab, with no matching of customers across systems
   2. BIdentity resolution, which matches records from the source systems to one unified individual profile
   3. CExporting every source table to a spreadsheet once a year, which the agent reads when a question arrives
   4. DTurning off the source systems so that all customer data is only ever read from the agent's own cache
   Show answer and explanationHide answer

   Correct answer: B

   Identity resolution matches records across sources into one unified profile that the agent can use. A dashboard does not unify records, a yearly spreadsheet is stale, and turning off source systems destroys the data.

5. Question 5Multi-Agent Interoperability

   An organization wants a sales agent to hand work to a separate service agent built by another team, with each agent keeping its own responsibilities. Which TWO practices support this design? (Choose two.)

   1. ADefine a clear contract for the handoff, including the inputs, outputs and the expected result of the work
   2. BGive one agent full access to the other agent's data and actions so that handoffs never fail for any reason
   3. CUse a documented interface, such as an API or an agent-to-agent protocol, for the communication between the agents
   4. DMerge both agents into one large agent so that there are no handoffs and no interfaces to maintain at all
   5. ELet each agent change the other's instructions at runtime so that the two agents always stay aligned with each other
   Show answer and explanationHide answer

   Correct answer: A, C

   A clear handoff contract and a documented interface keep the agents interoperable while each keeps its responsibilities. Full access to another agent's data, merging agents and runtime changes to each other's instructions break separation of responsibility and create unpredictable behaviour.