Act for many end users without crossing accounts
No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.
What counts as done
Every call runs on the connected account of the user in that session, and the log shows which user each call ran for; deleting a user removes their upstream tokens.
Measured, for our own tests: calls stay on the right user's account under concurrent load. No docs we read state these figures, so they don't count against a listing here.
How agents ask for it: run this for each customer
, use this user's CRM, not mine
Does the job, with a limit 6
It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.
| Listing | Why |
|---|---|
| Composio (API + MCP) A Sampark Inc (d/b/a Composio) | Each session is bound to one userId. acc-17 shows calls running as the last user to fetch tools (closed 28 October), acc-29 shows large ids colliding, and acc-30 shows upstream tokens kept after a user was deleted. source |
| MCP Gateway BB No legal entity found. COMMERCIAL.md, CLA.md and AUTHORSHIP.md name Mikko Parkkola, Finland, as licensor and author | End-user identity propagation (signed assertion, token passthrough or RFC 8693 exchange) fails closed, and per-user OAuth refuses calls without an identity. No per-user call log or token revocation on user deletion is documented. source |
| Merge Agent Handler BB Merge API, Inc. | A Registered User scopes every credential, tool call and audit record, and logs name the user. Deleting a user keeps credentials until the retention window ends and states no upstream token revocation. source |
| One B One Systems, Inc. | Each Connect user has their own grant, the Users tab shows reach and last use, and Activity lists calls. Revoking removes access in One, but upstream token revocation on deletion isn't stated. source |
| StackOne B StackOne Technologies Limited | Each request targets one linked account, projects isolate further, and account.deleted is an event. acc-26 showed concurrent SDK calls using the wrong account header (closed 27 April 2026), and upstream revocation on delete isn't stated. source |
| Unified.to MCP Server B Unified API Inc. | Each MCP URL is bound to one connection, linked to an end user by external_xref, so calls can't cross accounts. A per-user call log with arguments and upstream revocation on delete weren't found in the pages read. source |
Doesn't do this job 5
Listings in the category that do something else, with the reason.
| Listing | Why |
|---|---|
| AgentWeb D Agent Web Labs, Inc. | An aw_ key acts for one AgentWeb account and workspace; there is no end-user id per call for serving many customers from one process. source |
| letme BB ours Long Lunch Labs Limited (company number 17523112, registered in England and Wales, incorporated 15 October 2026), trading as Anchor Terminal and letme | letme has no end users or connections to keep apart; answers are the same for every caller. source |
| Monid C Monid, Inc. | One workspace balance and API keys; there are no end-user identities or per-user connections. source |
| Smithery F Clavia, Inc. (doing business as Smithery) | No hosted connections remain after the 31 October shutdown. source |
| Zapier MCP (agent actions) C Zapier Inc. | A server and its token belong to one Zapier user, and the docs say to give each user their own server and token; there is no per-call end-user parameter for one process serving many. source |
1 listings in the category were added after this reading and aren't marked yet.
What stops agents finishing
- SDK caches executors or headers across users
- user ids rounded or reused
Around it
Before: aggregator.connect-end-user. After: agent-observability.trace.
Where this comes from
Weight 11 of 100 in Agent tool access: 4 of 32 accounts (two SDK bugs that crossed users, one id collision, one stale-token case)
The same as data: /api/v1/use-cases/aggregator.isolate-users.json, or the MCP tool get_use_case with id aggregator.isolate-users. Other jobs in this category: /api/v1/niches?category=aggregator.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.