Agentforce “Answer Questions with Knowledge” Debugging

HomeMagulan Duraipandian

Agentforce “Answer Questions with Knowledge” Debugging

Magulan Duraipandian

Agentforce “Answer Questions with Knowledge” Troubleshooting: Fix the Field-Level Access Error

Your Agentforce Service Agent should answer from your Knowledge articles, but it returns an error instead. This guide covers the Answer Questions with Knowledge field-access error and gives you five checks to resolve it.

What is the error?

When the Answer Questions with Knowledge action runs against an Agentforce data library, you may see:

“Looks like you don’t have access to one or more fields used by the assigned data library. Update your field level access and try again.”

The agent’s running user can’t read one or more fields the data library depends on. The usual causes are missing permissions, missing field-level security (FLS), or a Data Cloud setup that isn’t complete.

Why does this happen?

An Agentforce Service Agent runs as a specific user, the user with the Einstein Agent User profile. Every object, field and Data Cloud resource the data library touches must be readable by that user. If any one of them isn’t, the action fails.

Quick checklist

#CheckWhere
1Data Cloud User permission set assignedAgent user
2Read access on Knowledge__kav and mapped fieldsPermission set
3Allow View Knowledge enabledPermission set or profile
4Data Space access verifiedAgent Studio and user permission set assignments
5Search index status is ReadyData Cloud > Search Index

Step 1: Assign the Data Cloud User permission set

If you use an Agentforce Service Agent, the user with the Einstein Agent User profile tied to the agent needs the Data Cloud User permission set assigned.

Step 2: Check object and field-level security (FLS)

  1. Go to Setup > Users > Permission Sets, or open the permission set assigned to your Agentforce service agent user.
  2. Click Object Settings and search for Knowledge (Knowledge__kav).
  3. Make sure Read access is enabled for the object.
  4. Open Field Permissions and confirm Read Access is enabled for every field selected in your Agentforce data library. That includes Title, Summary, Question and Details, plus any custom fields you mapped.

Tip: A single unreadable field can cause the error. Compare the fields selected in the data library with the field permissions line by line.

Step 3: Enable “Allow View Knowledge”

  1. In the same permission set, or in the profile or permissions assigned to the service agent, go to App Permissions or Knowledge Management.
  2. Enable Allow View Knowledge.

Step 4: Verify data space access

  1. Go to Setup > Agent Studio, select your Agent for Service, and view the assigned user details.
  2. Under Administration, click Users, select the user, and scroll to Permission Set Assignments.
  3. Confirm the user has the Data Cloud User permission set, or the appropriate Data Cloud Data Space Management permission set.
  4. Verify the data space is selected and checked.

Step 5: Confirm the search index status

  1. Go to Data Cloud > Search Index.
  2. Find the index for your Knowledge data library and confirm its status is Ready.

If the index is still building or has failed, the agent can’t pull articles.

Troubleshooting flow

Error appears
   ├─ Agent user has Data Cloud User permission set?  → No: assign it
   ├─ Read on Knowledge__kav + all mapped fields?     → No: fix FLS
   ├─ Allow View Knowledge enabled?                   → No: enable it
   ├─ Data space selected and checked?                → No: fix assignment
   └─ Search index = Ready?                           → No: wait or rebuild

Best practices

The material you shared contained no code snippet, so there were no code recommendations to extract. These best practices come from the configuration steps:

  1. Use a dedicated permission set for the agent user. Keep Knowledge read access, FLS and Data Cloud access in one named permission set. It’s easier to audit and reuse across sandboxes.
  2. Audit FLS whenever the data library changes. Adding or mapping a new field to the library, including custom fields, means granting Read access to the agent user at the same time.
  3. Grant least privilege. Give the agent user Read access only to the objects and fields the library needs.
  4. Check the search index status before deeper debugging. A status other than Ready blocks article retrieval, and the check takes a minute.
  5. Test as the agent user, not as an admin. Admin access can hide permission gaps.
  6. Re-verify after deployments. Permission sets, data space assignments and index status can differ between sandbox and production. [TBD: validate against your release process.]

FAQ

What does “Looks like you don’t have access to one or more fields used by the assigned data library” mean?
The Agentforce Service Agent user can’t read at least one field the data library uses. Review object permissions, FLS, the Allow View Knowledge permission and Data Cloud access.

Which user does the Agentforce Service Agent run as?
The user with the Einstein Agent User profile tied to the agent.

Which permission set does the Agentforce agent user need for Knowledge?
The Data Cloud User permission set, plus Read access on Knowledge__kav and the mapped fields.

Can an unready search index cause this issue?
An index that is still building or has failed prevents the agent from pulling articles. Confirm the status is Ready in Data Cloud > Search Index. The source material doesn’t say whether this specific message appears in that case, so check it as a separate cause.

Conclusion

Most cases of this error come down to one of five checks: the Data Cloud User permission set, Knowledge object and field permissions, Allow View Knowledge, data space access, and search index status. Work through them in order and re-test the agent after each fix.

Leave a Reply