n8n MCP Client Tool: Tools to Include modes for Sume

n8n's MCP Client Tool can expose All, Selected, or All Except tools. How to pick the set for Sume so an AI Agent node reads freely but cannot spend by accident.

5 min readSume
All posts

On n8n's MCP Client Tool node, set Tools to Include to Selected for Sume and tick the read tools plus only the one paid tool the workflow needs. The node offers three modes: All, Selected, and All Except. With a Sume API key the server lists its full hosted tool set, so All would hand an AI Agent every paid tool.

What are the three modes?

The n8n docs describe All as exposing every tool the server gives, Selected as choosing the tools to expose, and All Except as choosing the tools to avoid sharing. The node also supports Bearer, Generic header, multiple headers and OAuth2 authentication, or none.

Tools to Include modes applied to Sume (n8n docs read 2026-10-11)
ModeEffect on SumeWhen to use
AllAgent sees every tool the credential can seeRead-only OAuth session only
SelectedAgent sees only ticked toolsProduction workflows
All ExceptEverything but your exclusionsPrototyping with a key

Which Sume tools belong in a read-only set?

Sume's docs list the free reads: tools_list, tools_schema, mcp_health, account_me, balance_get, catalog_list, and the jobs reads jobs_list, jobs_get, jobs_status, jobs_result and jobs_wait. Assets reads are assets_list, assets_get and assets_download_url.

Paid examples are generate_image, generate_video, music_create, tts_create and avatars_create. An OAuth session without mcp:write never sees the write and paid tools, which is a second layer of control behind the n8n selection.

What else should the agent prompt say?

The n8n page labels the connection field as an SSE endpoint, while Sume documents a streamable HTTP endpoint. Test the connection with mcp_health before building around it, and see the related posts for how that field behaves.

  • Call tools_schema before the first paid call.
  • Send dry_run true first; it previews cost and does not submit.
  • Include a stable idempotency_key per workflow execution, and an optional max_spend_usd.
  • Use jobs_wait in slices of at most 55 seconds.

How should the workflow be wired?

Connect the MCP Client Tool to an AI Agent node, choose the authentication option that matches your credential (Bearer for a Sume API key, or OAuth2 if you have set up an OAuth client), and then set Tools to Include. Run the workflow once with a prompt that only asks the agent to list what it can do. The answer shows exactly which Sume tools survived your selection.

If a tool is missing, check two things in order: whether you ticked it in the n8n node, and whether the Sume credential can see it. A read-only OAuth session hides write and paid tools no matter what n8n selects. A key session shows them, so the n8n selection becomes the only filter, which is why Selected is the safer mode for anything that runs on a schedule.

Before you rely on any of this in a team setting, test it once end to end with a throwaway prompt. Call mcp_health, call tools_list, and run one dry_run against a paid tool. Compare the tool count with what you expected for your credential. If a read-only OAuth session lists more than you expected, or a key session lists fewer, stop and recheck which credential the client is actually sending. Sume's mcp_health response reports the auth source, which settles the question quickly without guessing.

Sources

Related posts

More in Integrations

All Integrations posts

Written by Sume