# Can an AI agent use Arcade? L3 · Does the work

Arcade is an MCP gateway for running tools such as calculators and apps on behalf of users, with a REST API and a CLI.

**L3 · Does the work** · [What the levels mean](https://agentdrives.com/method/levels?via=md)

Rank among the 246 products we drove:

- Core work done: 100%, tied #1 (8 of 8 core tasks)
- Time to get in: 4 minutes, tied #42 (Getting in on its own)
- Holds up: 75 / 100, tied #65 (Reliability under errors)
- Steps on its own: 3 / 5, tied #35 (Find to Leave, no person)

**Can an AI agent use Arcade? Yes.** Our agent signed up for Arcade on its own in a browser with an email and password, got an API key and used its API, MCP server and CLI.

- Level confirmed: Not yet (Level from one drive; a second must agree)
- Got in by: Sign-up in a browser (4 minutes)
- Got as far as: Pay: reaches the checkout (A card form for $25 a month)

Category: MCP gateways · Connects by: MCP, API, CLI · Agent payments: None found · Driven: 7 Nov 2026

**How to connect**
- MCP server URL: `https://api.arcade.dev/mcp/arcade` (streamable HTTP, documented). Source: https://api.arcade.dev/mcp/arcade
- API base URL: `https://api.arcade.dev` (documented). Source: https://www.arcade.dev/openapi.json
- Docs page: `https://docs.arcade.dev/en/home` (tested in our drive). Source: https://docs.arcade.dev/en/home

Checked 18 Nov 2026: MCP server asks for a sign-in; sign-up page, llms.txt and OpenAPI unchanged since the check before.

See [how it ranks in the MCP gateways category](https://agentdrives.com/categories/mcp-gateways?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. (322 doc pages read)
2. **Get in**: On its own. MCP sign-in, sign-up in a browser. (4 minutes)
3. **Use**: On its own. Did: search the tool catalog and fetch tool definitions in OpenAI and Anthropic formats; run a tool for a user and read the run back from the history. (8 of 8 core tasks)
4. **Pay**: Partly. Reaches the checkout. A card form for $25 a month. First stop.
5. **Leave**: Partly. Nothing to cancel. Its account stays.

To reach L4: Delete its account, so its keys and login stop working.

## How an agent connects

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

- **MCP server**: All tested. Official server. Site MCP (4 read-only tools, no auth); per-project gateways with OAuth (CIMD/DCR, PKCE), 7-9 tools when driven. Sign-in needed to see its tools. https://api.arcade.dev/mcp/arcade (10 tools mapped in our drive)
- **API**: 58 of 60 tested. REST at api.arcade.dev, OpenAPI 3.0.0. (60 operations mapped in our drive)
- **OpenAPI**: Yes. A machine-readable description of the API.
- **CLI**: 27 of 28 tested. arcade-mcp 1.17.3 on PyPI: login, new, deploy, server, secret, evals, and more. (28 commands mapped in our drive)
- **SDK**: None found.
- **llms.txt**: Yes. https://www.arcade.dev/llms.txt
- **Agent payments**: None found. Reached the card form and stopped: Growth, $25 a month. It checks out through a Stripe card form. No CAPTCHA at checkout.
- **Docs**: In part. Covers getting started, auth, error codes and code examples. Describes each operation only in part: completeness 50 of 100. 6 of 8 documented ways to do the core work worked. https://docs.arcade.dev/en/home (322 of 384 pages read)

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

## Getting in

Our agent tried 3 of 9 ways in; 3 worked without a person.

The way that worked: **Sign-up in a browser**, 4 minutes. Needs an inbox and a browser. No person needed.

- Agent on a server (API, MCP, CLI): No. Stops at sign-up: it works only in a browser.
- Assistant in its own browser (Like ChatGPT agent): Partly. Signs up, confirms its email and does all of the core work, then stops: its account stays. It can't delete its account.
- Signed agent (Known to Cloudflare): Yes. The sign-up page opens.

| Way in | What happened |
|---|---|
| Sign-up in a browser | Worked |
| MCP sign-in | Worked |
| Sign-up in a browser | Worked: Part of the work done |
| Runs locally, no account | Not tried: Not part of this drive |
| Continue with GitHub | Not tried: Not part of this drive |
| Continue with Google | Not tried: Not part of this drive |
| Continue with Microsoft | Not tried: Not part of this drive |
| Company single sign-on | Not tried: Needs a person's account |
| A key from a person | Not tried: The same as another way |

## What an agent can do with Arcade

All 8 core tasks done, and 15 of 20 others.

- Holds up: 75 of 100 (Reliability 83, Handles mistakes 67). The mean of the two.
- Docs match: 68 of 100 (Docs accuracy 75, Docs completeness 50). Docs accuracy counts 70%, completeness 30%.
- Surface tested: 88 of 100. 90 of 102 passed: the 105 documented operations less the 3 that spend money or reach real people

Core tasks:

| Task | Result | Done through |
|---|---|---|
| Run a tool action on behalf of a user | Done | API, MCP |
| Authorize a user's account for a tool (OAuth consent) | Done | API |
| Browse and search the tool catalog | Done | API, MCP |
| Create a tool gateway that federates selected tools | Done | API, MCP |
| Change or delete a tool gateway | Done | API |
| Build and deploy a custom tool server to Arcade | Done | API, CLI |
| Register a remote tool server in a project | Done | API |
| Add a contextual access hook to enforce policy on tool calls | Done | API |

15 of 20 other tasks done:

- Review tool execution history: Done
- Check a tool's unmet authorization requirements: Done
- Get tool definitions formatted for an LLM: Done
- Get a user's provider token to call a third-party API directly: Done
- Add a custom OAuth auth provider: Done
- Store a secret for tools: Done
- Revoke a user's connected account: Done
- Set user verification for the auth flow: Done
- Turn on tool recommendation for a gateway: Done
- Scaffold and run a tool server locally: Done
- Read a deployed server's logs: Done
- Answer end users' capability requests in the private registry: Done
- Read the audit log: Done
- Count tool calls for usage: Done
- Create an organization or project: Done
- Configure SSO for the Arcade organization: Paid plan
- Set a member's role (RBAC): Paid plan
- Schedule a tool execution for later: Blocked. The API refuses it as removed, although the OpenAPI still lists it.
- Run evaluations on tools: Blocked
- Call an LLM through Arcade: Blocked
- Connect an identity provider for gateway end users: Not tried
- Self-host Arcade in your own cloud with Helm: Not tried

The plans they need:

- Build and deploy a custom tool server to Arcade: paid (free plan allows one managed deployment at a time)

The drive used the free plan.

What it can't do, and why:

- Pay for the Growth plan: Only a card form in the dashboard; no API or agent payment option.
- Delete the account: No delete control or endpoint; the privacy policy says to email support.
- Schedule a tool run for later: The API refuses it as removed, although the OpenAPI still lists it.
- Call an LLM through Arcade: The documented chat endpoint answers 404.
- Deploy more than one hosted server at a time: The free plan refuses a second deployment.
- Finish OAuth for prebuilt GitHub and Google tools: Consent needs a person's GitHub or Google account.

**Surface audit:** 105 documented operations: 99 run, 3 blocked by sandbox or plan limits and 3 left out because they spend money or reach real people. Of the 99 run, 90 passed and 9 didn't; 25 behaved differently from the docs. By surface, 58 on the API, 27 in the CLI, 10 on the MCP server and 4 elsewhere, such as sign-in and webhooks.

## Tips for agent builders

- After sign-up, set up the authenticator-app second factor: the setup key is shown as text, so you can compute TOTP codes yourself.
- Prefer the MCP gateway at https://api.arcade.dev/mcp/arcade: it supports CIMD and PKCE and needs one consent click.
- Use the org-scoped paths /v1/orgs/{org}/projects/{project}/plugins and /hooks: the flat /v1/plugins and /v1/hooks in openapi.json answer 404.
- Run `arcade deploy -e src/<pkg>/server.py` from the package root, not from the src folder the scaffold's next steps point to.
- Delete a user connection with its connection_id (ac_...), not the list item's id (at_...).
- Read tool_executions with the te_ id from the execute response; the tc_ execution_id is refused.
- When authorizing for a user_id, sign in to the consent browser as that same Arcade user: the default user verification refuses others with user_mismatch.
- Custom OAuth providers need authorize_request.params with {{client_id}} and {{redirect_uri}}, or the consent fails with unknown_client.

## Paying

Ways to pay checked 7 Nov 2026.

- Result: Partly. Reaches the checkout. A card form for $25 a month.
- Checkout: Reached the card form and stopped: Growth, $25 a month. It checks out through a Stripe card form. No CAPTCHA at checkout.
- Trial: No card needed to start.

## Leaving

- Result: Partly. Nothing to cancel. Its account stays.
- Its account: Not on its own: blocked. The account stays. Manage my account and the dashboard have no delete control, and the privacy policy says to email support@arcade.dev; mailing a real support desk is outside our rules. Everything we made was removed and both API keys refused afterwards (401), but the login still works.

## Checked every day

What our agent finds on the product's own site each day, without an account.

- Sign-up page, loads on the newest check; 11 of 11 days
- llms.txt, found on the newest check; 11 of 11 days
- MCP server, asks for a sign-in on the newest check; 4 of 11 days
- OpenAPI, found on the newest check; 11 of 11 days

## How Arcade compares

In MCP gateways: None of the 3 products we drove in MCP gateways is higher; 2 others share L3. As of 7 Nov 2026.

- **Arcade** (this page): L3, Gets in on its own
- [Nango](https://agentdrives.com/tools/nango.md?via=md): L3, Gets in on its own
- [Composio](https://agentdrives.com/tools/composio.md?via=md): L3, Gets in on its own

[All 3 in MCP gateways](https://agentdrives.com/categories/mcp-gateways?via=md)

[How we rank](https://agentdrives.com/method?via=md#order)

On our standard (34 criteria, each against every product we measured on it). Best or ahead on 8 of 34 criteria, behind on 6.

- [Agents pick the right tool](https://agentdrives.com/standard#mcp-05): **Ahead of 73 of 91** others, 18 the same.
- [llms.txt that works](https://agentdrives.com/standard#find-02): **Ahead of 185 of 239** others, 54 the same.
- [A path written for agents](https://agentdrives.com/standard#docs-10): **Ahead of 140 of 187** others, 47 the same.
- [Every way in works for an agent](https://agentdrives.com/standard#in-11): **Ahead of 134 of 193** others, 59 the same.
- [A way in with no person](https://agentdrives.com/standard#in-01): **Ahead of 137 of 243** others, 106 the same.
- [Truthful annotations](https://agentdrives.com/standard#mcp-07): **Ahead of 50 of 91** others, 41 the same.
- [A tool list that fits](https://agentdrives.com/standard#mcp-11): **Ahead of 48 of 91** others, 43 the same.
- [Sign-in for agents acting for a person](https://agentdrives.com/standard#in-08): **Level with 83 of 158** others, 75 lower.
- [All validation problems at once](https://agentdrives.com/standard#err-04): **Ahead of 34 of 57** others, 12 the same, 11 higher.
- [One way to page through lists](https://agentdrives.com/standard#api-03): **In the middle** 49 of 114 others lower, 52 the same, 13 higher.
- [Prices an agent can read](https://agentdrives.com/standard#find-06): **Level with 117 of 153** others, 36 lower.
- [Agents may read docs, API and sign-up](https://agentdrives.com/standard#find-01): **Level with 129 of 160** others, 31 lower.
- [Fast to the first call](https://agentdrives.com/standard#in-03): **Level with 118 of 135** others, 17 lower.
- [Descriptions that say when to use the tool](https://agentdrives.com/standard#mcp-04): **Level with 56 of 91** others, 22 lower, 13 higher.
- [The best way in is built for agents](https://agentdrives.com/standard#in-12): **In the middle** 79 of 196 others lower, 55 the same, 62 higher.
- [Terms allow an agent's account](https://agentdrives.com/standard#in-10): **Level with 179 of 189** others, 10 lower.
- [Real status codes](https://agentdrives.com/standard#err-01): **Level with 161 of 168** others, 7 lower.
- [Errors say what to do next](https://agentdrives.com/standard#err-03): **Level with 107 of 145** others, 22 lower, 16 higher.
- [Typed, tight input schemas](https://agentdrives.com/standard#mcp-06): **Level with 59 of 91** others, 17 lower, 15 higher.
- [WebMCP tools on the web app](https://agentdrives.com/standard#proto-02): **Level with 133 of 136** others, 3 higher.
- [Structured results](https://agentdrives.com/standard#mcp-08): **In the middle** 22 of 78 others lower, 32 the same, 24 higher.
- [An agent payment rail on the core](https://agentdrives.com/standard#pay-02): **Level with 205 of 216** others, 11 higher.
- [A request id on every answer](https://agentdrives.com/standard#err-07): **In the middle** 42 of 143 others lower, 47 the same, 54 higher.
- [Results sized for a context window](https://agentdrives.com/standard#mcp-09): **In the middle** 16 of 70 others lower, 27 the same, 27 higher.
- [Docs an agent can navigate](https://agentdrives.com/standard#docs-11): **Level with 108 of 187** others, 19 lower, 60 higher.
- [Spend an agent can see and cap](https://agentdrives.com/standard#pay-05): **Level with 105 of 159** others, 54 higher.
- [Pages fit one fetch](https://agentdrives.com/standard#docs-08): **Level with 118 of 187** others, 69 higher.
- [Limits known before they hit](https://agentdrives.com/standard#lim-02): **Level with 47 of 90** others, 43 higher.
- [Versions an agent can pin](https://agentdrives.com/standard#api-07): **Behind 74 of 127** others, 43 the same, 10 lower.
- [Delete by API, checked](https://agentdrives.com/standard#leave-02): **Behind 52 of 97** others, 45 the same.
- [Answers sized for an agent](https://agentdrives.com/standard#api-04): **Behind 86 of 116** others, 13 the same, 17 lower.
- [Export before leaving](https://agentdrives.com/standard#leave-03): **Behind 117 of 189** others, 72 the same.
- [Errors a program can read](https://agentdrives.com/standard#err-02): **Behind 99 of 145** others, 46 the same.
- [Tool names that sort themselves](https://agentdrives.com/standard#mcp-03): **Behind 86 of 91** others, 5 the same.

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

## Agents get stuck at Pay

**7** things to fix for agents: 4 small, 2 medium and 1 large

One fix can cover several findings.

Agent blockers: **6 found** Agent surfaces: **27 issues found**

[Get your findings, free](https://agentdrives.com/your-findings?p=arcade) · [What a drive covers](https://agentdrives.com/method?via=md)

We mail a private link to an address at arcade.dev.

1. Find Clear
2. Get in Clear
3. Use Clear
4. Pay Blocks agents
5. Leave Partly

Or [claim Arcade](https://agentdrives.com/go/claim/arcade?add=arcade.dev&ref=listing) to see where each issue is and ask for a free drive after a fix.

Want Arcade driven every month? [See Watch](https://agentdrives.com/owners#watch)

### On our standard

Best or ahead on 8 of 34 criteria, behind on 6. Each against every product we measured on it.

- [Agents pick the right tool](https://agentdrives.com/standard#mcp-05)
- [llms.txt that works](https://agentdrives.com/standard#find-02)
- [A path written for agents](https://agentdrives.com/standard#docs-10)
- [Every way in works for an agent](https://agentdrives.com/standard#in-11)
- [Answers sized for an agent](https://agentdrives.com/standard#api-04)
- [Export before leaving](https://agentdrives.com/standard#leave-03)
- [Errors a program can read](https://agentdrives.com/standard#err-02)
- [Tool names that sort themselves](https://agentdrives.com/standard#mcp-03)

**All 34 criteria**

- [A way in with no person](https://agentdrives.com/standard#in-01)
- [Truthful annotations](https://agentdrives.com/standard#mcp-07)
- [A tool list that fits](https://agentdrives.com/standard#mcp-11)
- [Sign-in for agents acting for a person](https://agentdrives.com/standard#in-08)
- [All validation problems at once](https://agentdrives.com/standard#err-04)
- [One way to page through lists](https://agentdrives.com/standard#api-03)
- [Prices an agent can read](https://agentdrives.com/standard#find-06)
- [Agents may read docs, API and sign-up](https://agentdrives.com/standard#find-01)
- [Fast to the first call](https://agentdrives.com/standard#in-03)
- [Descriptions that say when to use the tool](https://agentdrives.com/standard#mcp-04)
- [The best way in is built for agents](https://agentdrives.com/standard#in-12)
- [Terms allow an agent's account](https://agentdrives.com/standard#in-10)
- [Real status codes](https://agentdrives.com/standard#err-01)
- [Errors say what to do next](https://agentdrives.com/standard#err-03)
- [Typed, tight input schemas](https://agentdrives.com/standard#mcp-06)
- [WebMCP tools on the web app](https://agentdrives.com/standard#proto-02)
- [Structured results](https://agentdrives.com/standard#mcp-08)
- [An agent payment rail on the core](https://agentdrives.com/standard#pay-02)
- [A request id on every answer](https://agentdrives.com/standard#err-07)
- [Results sized for a context window](https://agentdrives.com/standard#mcp-09)
- [Docs an agent can navigate](https://agentdrives.com/standard#docs-11)
- [Spend an agent can see and cap](https://agentdrives.com/standard#pay-05)
- [Pages fit one fetch](https://agentdrives.com/standard#docs-08)
- [Limits known before they hit](https://agentdrives.com/standard#lim-02)
- [Versions an agent can pin](https://agentdrives.com/standard#api-07)
- [Delete by API, checked](https://agentdrives.com/standard#leave-02)

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

## For the team behind Arcade

**7 things to fix for agents: 4 small, 2 medium and 1 large.**

One fix can cover several findings. Agent blockers: 6 found · Agent surfaces: 27 issues found.

Agents get stuck at Pay. Its team gets the two findings that matter most in full, free, by a private link to an address at arcade.dev.

- Find: Clear
- Get in: Clear
- Use: Clear
- Pay: Blocks agents
- Leave: Partly

[Get your findings, free](https://agentdrives.com/your-findings?p=arcade) · [Claim Arcade](https://agentdrives.com/app/products?add=arcade.dev&ref=listing) · [What a drive covers](https://agentdrives.com/method?via=md)

## Read more on this step

Pay is where our agent stopped at Arcade.

- [How far does an AI agent get in a card checkout?](https://agentdrives.com/findings/card-checkout?via=md) (finding): Our agent reached the checkout at 88 of the 190 products it set out to buy from, and the card form at 68.
- [Let AI agents pay you per call: x402 and MPP](https://agentdrives.com/guides/let-agents-pay-per-call?via=md) (guide): x402 or MPP on a paid endpoint, your prices where agents read them, and plans an agent can buy by API or MCP.
- [Tell an AI agent a feature needs a paid plan](https://agentdrives.com/guides/tell-ai-agents-a-feature-needs-a-paid-plan?via=md) (guide): How to answer an AI agent at a paid-plan wall: 402 or 403, a Problem Details body with the required plan, and an upgrade link a person can open.

[Everything on Pay](https://agentdrives.com/access/pay)

## Questions people ask about Arcade

**Can an AI agent create an Arcade account on its own?**

Yes, in a browser, without a person. Done by the agent in its own browser at app.arcade.dev/register (account.arcade.dev): email, company, full name, nickname and password, no CAPTCHA (company and name asked; our agent answered). After email verification the dashboard requires a passkey or 2FA. The agent set up an authenticator app: the setup key is shown as text, so it computed the codes itself.

**Does Arcade have an MCP server, and can an agent connect to it?**

Yes. Arcade has an official MCP server. The server signs an agent in with OAuth. A new agent can get in on its own: in the drive of 7 Nov 2026, our agent signed up, verified and got its key without a person. Through the MCP server, our agent tried 3 tasks and did 3.

**How does an AI agent get an Arcade API key?**

In the drive of 7 Nov 2026, our agent got its Arcade API key in a browser, without a person. The agent made a project API key on the dashboard's API Keys page in its own browser (the documented route). MCP sign-in (CIMD, PKCE) and `arcade login` gave OAuth tokens through consent in the agent's browser. Contrary to the docs, `arcade login` stores a user token, not an API key.

**Can an AI agent delete its Arcade account?**

Not on its own: blocked. The account stays. Manage my account and the dashboard have no delete control, and the privacy policy says to email support@arcade.dev; mailing a real support desk is outside our rules. Everything we made was removed and both API keys refused afterwards (401), but the login still works.

**Can an AI agent pay for Arcade?**

Partly. Our agent reached a card form for $25 a month. No card needed to start.

## Where Arcade fits

- Category: [AI and agent infrastructure](https://agentdrives.com/categories/ai?via=md), [MCP gateways](https://agentdrives.com/categories/mcp-gateways?via=md)
- Tasks: [Read usage and quotas](https://agentdrives.com/tasks/read-usage-and-quotas), [Read the audit log](https://agentdrives.com/tasks/read-the-audit-log), [Create a workspace](https://agentdrives.com/tasks/create-a-workspace), [Register an OAuth app](https://agentdrives.com/tasks/register-an-oauth-app), [Read logs](https://agentdrives.com/tasks/read-logs), [Set environment variables and secrets](https://agentdrives.com/tasks/set-environment-variables-and-secrets)
- Wall: [Let AI agents start without a credit card](https://agentdrives.com/guides/card-free-starts?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) (first stop), [Leave](https://agentdrives.com/access/leave)
- Topics: [MCP servers](https://agentdrives.com/access/find/mcp-servers) (Find step), [Docs and llms.txt](https://agentdrives.com/access/find/docs) (Find step), [MCP sign-in](https://agentdrives.com/access/get-in/mcp-sign-in) (Get in step), [Sign-up](https://agentdrives.com/access/get-in/sign-up) (Get in step), [MCP and API](https://agentdrives.com/access/use/mcp-and-api) (Use step), [Card checkout](https://agentdrives.com/access/pay/card-checkout) (Pay step), [Delete the account](https://agentdrives.com/access/leave/delete-account) (Leave step)
- Rivals: [Composio](https://agentdrives.com/tools/composio?via=md) (L3), [Nango](https://agentdrives.com/tools/nango?via=md) (L3)

The level follows [these rules](https://agentdrives.com/changes#rules-2026-11-18).

## Level by drive

One drive so far, on 7 Nov 2026.

[Every drive in full](https://agentdrives.com/tools/arcade/history?via=md)

Source: our drives of Arcade. As of 7 Nov 2026.

1. 11 Nov 2026 [**Level L4 → L3**](https://agentdrives.com/tools/arcade?via=md#drive-20261107T083611Z) [Re-scored under method v47, drive of 7 Nov 2026](https://agentdrives.com/changes#rules-2026-11-11-47)

Data: [JSON](https://agentdrives.com/api/tools/arcade.json?via=md) · [Markdown](https://agentdrives.com/tools/arcade.md?via=md) · [History](https://agentdrives.com/tools/arcade/history?via=md) · [Atom](https://agentdrives.com/feeds/tools/arcade.atom) · [JSON Feed](https://agentdrives.com/feeds/tools/arcade.json) · licence [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) · evidence sha256 `3ac99f9104b646a63e992c6d67e28d58882e40a3b6516f73511f3f27af0dab29`

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