# Clerk L4 · Runs on its own

**Can an AI agent use Clerk? Yes, without signing up.** Our agent got into Clerk on its own in 3 minutes and did all 6 of its core tasks.

Category: Auth and identity · Connects by: MCP, API, CLI, SDK · Agent payments: None found · Driven: 1 Nov 2026

**How to connect**
- MCP server URL: `https://mcp.clerk.com/mcp` (streamable HTTP, documented). Source: https://mcp.clerk.com/mcp
- API base URL: `https://api.clerk.com/v1` (documented). Source: https://clerk.com/openapi.json
- CLI install: `npm install -g clerk` (documented). Source: https://clerk.com/llms.txt
- Docs page: `https://clerk.com/docs` (tested in our drive). Source: https://clerk.com/docs

**Level 4: Runs on its own.** An agent does the core work and can stop paying on its own.
- Got in by: No key needed (3 minutes, no account)
- Got as far as: Use: all 6 core tasks (No payment made)
- To reach L5: Pay without a person.

2 drives on one day, awaiting confirmation

[What the levels mean](https://agentdrives.com/method/levels?via=md)

## From finding it to leaving

What an agent managed without a person, at each step.

1. **Find**: On its own. Docs it can act on: llms.txt, an OpenAPI and an MCP server. (599 doc pages read)
2. **Get in**: On its own. Got in without signing up. (3 minutes)
3. **Use**: On its own. Did: create, read, update and delete users, and sign them in; create organizations, add members and change their roles. (6 of 6 core tasks)
4. **Pay**: Not reached. Paying needs an account.
5. **Leave**: On its own. Nothing to cancel; what it made stays.

## Getting in

2 of 9 ways in worked without a person.

The way that worked: **No key needed**, 3 minutes. No inbox, browser or person needed.

- Agent on a server (API, MCP, CLI): Partly. Without an account (no key needed) it gets an API key, makes its first API call and does all of the core work. Payment wasn't tried.
- Assistant in its own browser (Like ChatGPT agent): No. Signing up needs a person: the form refused our agent's browser.
- Signed agent (Known to Cloudflare): Yes. The sign-up page opens.

| Way in | What happened |
|---|---|
| No key needed | Worked |
| Try first, claim later | Worked |
| Email and password | Refused |
| CLI code login | Stopped: The consent page needs an account |
| Sign-in for apps (OAuth, MCP) | Stopped: The consent page needs an account |
| Continue with GitHub | Not tried: Not part of this drive |
| Continue with Google | Not tried: Not part of this drive |
| Company single sign-on | Not tried: Needs a person's account |
| Added by an admin | Not tried: Needs a person's account |

[How agents get past bot checks](https://agentdrives.com/guides/let-signed-agents-past-your-bot-check.md?via=md)

## What an agent can do with Clerk

All 6 core tasks done, and 10 of 21 others.

- Reliability: core tasks that worked on all 3 runs: 69 of 100
- Handles mistakes: clear errors, retries without duplicates, honest limits: 52 of 100
- Surface tested: documented routes, tools and commands run and checked: 59 of 100
- Docs accuracy: documented operations that behaved as the docs say: 99 of 100
- Docs completeness: how fully the docs describe each operation: 53 of 100

Core tasks:

| Task | Result | Done through | Offered, not tried |
|---|---|---|---|
| Authenticate your app's end users with Clerk's components and API | Done | API, CLI | MCP |
| Configure sign-in methods for your app | Done | API | MCP |
| Create and manage organizations for a multi-tenant app | Done | API | – |
| Set a member's role or permissions | Done | API | – |
| Set up enterprise SSO (SAML, OIDC, EASIE) | Done | API | MCP |
| Start a subscription for your app's customer | Done | API | – |

10 of 21 other tasks done:

- Style and brand the prebuilt sign-in UI: Done
- Invite a member to an organization: Done
- Provision and deprovision users with SCIM Directory Sync: Done
- Create an API key for machine-to-machine auth: Done
- Set a price for a billing plan: Done
- Set up a self-service billing portal for customers: Done
- Ban or block a user by attribute: Done
- Impersonate a user for support troubleshooting: Done
- Read application and admin activity logs: Done
- Collect sign-ups in a waitlist before launch: Done
- Register a webhook for Clerk events: Partly
- Export all user data: Partly
- Connect a custom domain: Partly
- Set a custom session duration and track devices: Blocked. Those settings need a claimed app with an account.
- Require multi-factor authentication for end users: Blocked
- Restrict Secret Key access by IP address: Blocked
- Change the Clerk subscription plan: Blocked. Those settings need a claimed app with an account.
- Remove Clerk branding from prebuilt UI: Blocked
- Download the account's invoices: Blocked
- Read usage and quota consumption: Blocked
- Provision a Clerk app for a customer via the Platform API: Blocked

What it can't do, and why:

- Make a Clerk Dashboard account: Sign-up refused our agent's browser twice with no challenge shown.
- Change sign-in methods, MFA or session length: Those settings need a claimed app with an account.
- Use Clerk's MCP server: Its sign-in needs a Clerk account.
- Buy a Clerk plan: Only the Dashboard sells plans; no billing API found.
- Delete the two unclaimed apps: The docs we read name no delete or expiry without claiming them.

**Surface audit:** 636 documented operations, 385 run, 349 passed, 5 not as documented, 206 blocked (sandbox or plan limits), 45 left out to stay safe.

## How an agent connects

MCP, API, OpenAPI, CLI, SDK and llms.txt found.

Agent surfaces: API tested · MCP documented · CLI tested · SDK tested · Docs tested · Agent payments none found. Issues found: 6.

- **MCP server**: Listed, not tested. Official server. Remote server, 21 tools when driven in the docs, OAuth sign-in. https://mcp.clerk.com/mcp (21 tools across its docs) 20 tools listed without signing in on 30 Oct 2026: `list_applications`, `get_application`, `create_application`, `get_instance_config`, `get_instance_config_schema`, `patch_instance_config`, `put_instance_config`, `list_users`, `create_user`, `list_domains`, `get_domain_status`, `check_domain_dns`, `get_deploy_oauth_providers`, `create_production_instance`, `set_oauth_provider_credentials`, `list_application_feedback`, `submit_feedback`, `get_more_tools`, `clerk_sdk_snippet`, `list_clerk_sdk_snippets`.
- **API**: 349 tested. REST Backend API, 263 operations (the 20 Aug 2026 MCP spec); Frontend API 214; Platform API 70 (BETA) (549 operations across its docs)
- **OpenAPI**: Yes. 3.0.3 for all three specs.
- **CLI**: 20 tested. clerk 4.1.0 (npm), agent mode, accountless init. (45 commands across its docs)
- **SDK**: All tested. Packages: PyPI [clerk-backend-api](https://pypi.org/project/clerk-backend-api/). (10 methods across its docs)
- **llms.txt**: Yes. https://clerk.com/llms.txt
- **Agent payments**: None found. The checkout wasn't reached: Pro, $25 a month.
- **Docs**: No gaps in what we read. Covers getting started, auth, error codes and code examples. 4 of 10 documented ways to do the core work worked when tried. https://clerk.com/docs (599 of 2,796 pages read)

[The standards](https://agentdrives.com/standard)

## Tips for agent builders

- Skip the Dashboard: POST https://api.clerk.com/v1/accountless_applications (no auth) returns a full-scope development secret key; or run `npx -y clerk@latest init --accountless` in a supported framework folder.
- Use +clerk_test addresses on example.com for test users: no mail is sent and code 424242 passes email verification and second-factor checks.
- Run the CLI with `--mode agent` and close stdin (`</dev/null`): `clerk api … -X DELETE` waits on an open stdin.
- Change user metadata through PATCH /v1/users/{id}/metadata: public_metadata on PATCH /v1/users/{id} is refused as deprecated.
- Start or change a customer's subscription with POST /v1/billing/subscription_items/{id}/price_transition (a custom $0 price works on development); there is no POST /v1/billing/subscriptions.
- Pass permission ids (perm_…), not keys, when creating organization roles.
- Browser-free sign-ups through the Frontend API can be refused as suspicious; create users with POST /v1/users instead.

## How Clerk compares

On our standard (21 criteria, each against every product we measured on it):

- [Errors a program can read](https://agentdrives.com/standard#err-02): **Ahead of most** 84 of 104 others lower, 20 the same.
- [Docs an agent can navigate](https://agentdrives.com/standard#docs-11): **Ahead of most** 98 of 141 others lower, 43 the same.
- [Spend an agent can see and cap](https://agentdrives.com/standard#pay-05): **Ahead of most** 85 of 118 others lower, 27 the same, 6 higher.
- [The best way in is built for agents](https://agentdrives.com/standard#in-12): **Ahead of most** 98 of 147 others lower, 49 the same.
- [A person can claim what the agent made](https://agentdrives.com/standard#in-09): **Ahead of most** 5 of 9 others lower, 4 the same.
- [Sign-in for agents acting for a person](https://agentdrives.com/standard#in-08): **Level with most** 64 of 119 others the same, 55 lower.
- [Versions an agent can pin](https://agentdrives.com/standard#api-07): **Level with most** 59 of 97 others the same, 38 lower.
- [One way to page through lists](https://agentdrives.com/standard#api-03): **In the middle** 35 of 87 others lower, 42 the same, 10 higher.
- [Answers sized for an agent](https://agentdrives.com/standard#api-04): **Level with most** 61 of 85 others the same, 24 lower.
- [Prices an agent can read](https://agentdrives.com/standard#find-06): **Level with most** 80 of 105 others the same, 25 lower.
- [Export before leaving](https://agentdrives.com/standard#leave-03): **In the middle** 56 of 144 others lower, 60 the same, 28 higher.
- [Fast to the first call](https://agentdrives.com/standard#in-03): **Level with most** 71 of 85 others the same, 14 lower.
- [Every way in works for an agent](https://agentdrives.com/standard#in-11): **In the middle** 56 of 145 others lower, 51 the same, 38 higher.
- [A path written for agents](https://agentdrives.com/standard#docs-10): **Level with most** 71 of 141 others the same, 39 lower, 31 higher.
- [Real status codes](https://agentdrives.com/standard#err-01): **Level with most** 125 of 127 others the same, 2 lower.
- [Errors say what to do next](https://agentdrives.com/standard#err-03): **Level with most** 56 of 103 others the same, 24 lower, 23 higher.
- [WebMCP tools on the web app](https://agentdrives.com/standard#proto-02): **Level with most** 91 of 93 others the same, 2 higher.
- [A request id on every answer](https://agentdrives.com/standard#err-07): **In the middle** 34 of 103 others lower, 29 the same, 40 higher.
- [All validation problems at once](https://agentdrives.com/standard#err-04): **Level with most** 23 of 35 others the same, 12 higher.
- [Pages fit one fetch](https://agentdrives.com/standard#docs-08): **Level with most** 87 of 141 others the same, 54 higher.
- [Limits known before they hit](https://agentdrives.com/standard#lim-02): **Level with most** 26 of 47 others the same, 21 higher.

[How each criterion is graded](https://agentdrives.com/standard)

## For the team behind Clerk

**7 things to fix for agents.**

Agent blockers: 1 found · Agent surfaces: 5 issues found. Safety review: 2 issues found.

Our agent didn't reach Pay. Prove Clerk is yours, then see where each issue is, read your safety findings and ask for a free drive after a fix. How to fix each one comes with the Audit.

- Find: Clear
- Get in: Clear
- Use: Clear
- Pay: Not reached
- Leave: Clear

[Claim Clerk, free](https://agentdrives.com/app/products?add=clerk.com&ref=listing) · [What a drive covers](https://agentdrives.com/method?via=md)

## Where Clerk fits

- Category: [Developer platform](https://agentdrives.com/categories/developer?via=md), [Auth and identity](https://agentdrives.com/categories/auth?via=md)
- Tasks: [Invite a member](https://agentdrives.com/tasks/invite-a-member), [Create an API key](https://agentdrives.com/tasks/create-an-api-key), [Set a member's role or permissions](https://agentdrives.com/tasks/set-a-member-s-role-or-permissions), [Read the audit log](https://agentdrives.com/tasks/read-the-audit-log), [Set the brand logo and colours](https://agentdrives.com/tasks/set-the-brand-logo-and-colours), [Sign up and sign in your app's users](https://agentdrives.com/tasks#sign-up-and-sign-in-your-app-s-users)
- Wall: [Bot checks](https://agentdrives.com/access/get-in/bot-checks) (our agent got in another way), [Let AI agents delete the account they opened](https://agentdrives.com/guides/let-agents-delete-their-account?via=md) (fix guide)
- Standards: [MCP](https://agentdrives.com/standards/mcp) (tools an agent can call), [OpenAPI](https://agentdrives.com/standards/openapi) (its API, described), [llms.txt](https://agentdrives.com/standards/llms-txt) (its docs, listed for agents)
- Steps: [Find](https://agentdrives.com/access/find), [Get in](https://agentdrives.com/access/get-in), [Use](https://agentdrives.com/access/use), [Pay](https://agentdrives.com/access/pay), [Leave](https://agentdrives.com/access/leave)

Data: [JSON](https://agentdrives.com/api/tools/clerk.json?via=md) · [Markdown](https://agentdrives.com/tools/clerk.md?via=md) · [History](https://agentdrives.com/tools/clerk/history?via=md) · [Atom](https://agentdrives.com/feeds/tools/clerk.atom) · [JSON Feed](https://agentdrives.com/feeds/tools/clerk.json) · licence [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)

**Found an error?** [Tell us on this page](https://agentdrives.com/tools/clerk?via=md#correct) or read [how corrections work](https://agentdrives.com/corrections?via=md).
