# Couchbase L4 · Runs on its own

Couchbase Capella is a cloud database.

**Can an AI agent use Couchbase? Yes, in a browser.** Our agent got into Couchbase on its own in 8 minutes and did all 8 of its core tasks.

Category: Databases · Connects by: MCP, API, CLI, SDK · Agent payments: None found · Driven: 25 Oct 2026

**How to connect**
- API base URL: `https://cloudapi.cloud.couchbase.com` (documented). Source: https://docs.couchbase.com/cloud/management-api-reference/_attachments/openapi.external.generated.yaml
- Docs page: `https://docs.couchbase.com/home/index.html` (tested in our drive). Source: https://docs.couchbase.com/home/index.html

**Level 4: Runs on its own.** An agent does the core work and can stop paying on its own.
- Got in by: Email and password (8 minutes, in a browser)
- Got as far as: Pay: reaches the checkout (Stopped at the card form)
- To reach L5: Pay without a person.

1 drive, 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. (462 doc pages read)
2. **Get in**: On its own. Sign-up in a browser. (8 minutes)
3. **Use**: On its own. Did: create and delete a free database cluster; add and read documents, run queries and search, including vector search. (8 of 8 core tasks)
4. **Pay**: Reaches the checkout. Stopped at the card form.
5. **Leave**: On its own. Nothing to cancel, deletes its account.

## Getting in

1 of 5 ways in worked without a person.

The way that worked: **Email and password**, 8 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. Payment: reaches the checkout, stopped at the card form.
- Signed agent (Known to Cloudflare): Yes. The sign-up page opens.

| Way in | What happened |
|---|---|
| Email and password | Worked |
| 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 |
| Added by an admin | Not tried: Needs a person's account |

## What an agent can do with Couchbase

All 8 core tasks done, and 8 of 10 others.

- Reliability: core tasks that worked on all 3 runs: 100 of 100
- Handles mistakes: clear errors, retries without duplicates, honest limits: 94 of 100
- Surface tested: documented routes, tools and commands run and checked: 54 of 100
- Docs accuracy: documented operations that behaved as the docs say: 90 of 100
- Docs completeness: how fully the docs describe each operation: 86 of 100

Core tasks:

| Task | Result | Done through | Offered, not tried |
|---|---|---|---|
| Create a database cluster | Done | API | CLI |
| Create or update a document (key-value) | Done | API, MCP | CLI |
| Delete a document | Done | API, MCP | CLI |
| Run a SQL++ (N1QL) query over a database | Done | API, MCP, CLI | – |
| Run a full-text search query | Done | API | – |
| Run a vector search | Done | API | – |
| List databases/buckets and read their schema | Done | API, MCP | CLI |
| Sync a local/mobile database with the cloud | Done | API | – |

8 of 10 other tasks done:

- Create a table/collection: Done
- Set row-level security / RBAC roles for users: Done
- Back up a database: Done
- Find slow or inefficient queries: Done
- Add knowledge/tools to an AI agent via Agent Catalog: Done
- Invite a member / manage organization users: Done
- Create an API key: Done
- Read usage and quotas / billing: Done
- Restore a database from backup: Partly
- Export data to a warehouse / connector: Partly
- Set up cross data-center replication (XDCR): Paid plan
- Create a database trigger / eventing function: Paid plan. These need a paid plan.
- Run an analytics (OLAP) query: Paid plan. These need a paid plan.
- Resize/scale database compute: Paid plan
- Add a read replica / server group: Paid plan
- Deploy a hosted AI agent / model: Paid plan
- Encrypt data with a customer-managed key: Paid plan
- Configure SAML/LDAP SSO: Paid plan
- Read the audit log: Paid plan

The plans they need:

- Back up a database (managed backups): paid
- XDCR: paid
- Eventing: paid
- Analytics: paid
- Resize: paid
- Replicas / server groups: paid
- Hosted model: paid
- CMEK: paid
- SSO: paid
- Audit log: Enterprise

Paid-plan tasks don't count toward the level. The drive used the free plan.

What it can't do, and why:

- Get the first API key without a browser: The first key can only be made in the web page.
- Upgrade to a paid plan: Payment is a card typed into an in-app form.
- Use backups service, analytics, eventing and scaling on the free tier: These need a paid plan.
- Read the pricing page: Cause not confirmed in this drive.

**Surface audit:** 426 documented operations, 230 run, 206 passed, 24 not as documented, 150 blocked (sandbox or plan limits), 4 not run for time, 45 left out to stay safe.

## How an agent connects

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

Agent surfaces: API tested · MCP tested · CLI tested · SDK tested · Docs tested · Agent payments none found. Issues found: 29.

- **MCP server**: All tested. Official server. Self-hosted, 33 tools. (24 tools across its docs)
- **API**: 168 tested. REST Management API v4, 343 operations; Data API, 18 operations. (361 operations across its docs)
- **OpenAPI**: Yes. 3.0.2.
- **CLI**: 20 tested. cbsh 1.4.0; cbq, cbimport, cbexport, cbbackupmgr 8.0.3; agentc 1.1.5. (22 commands across its docs)
- **SDK**: All tested. Node.js couchbase 4.7.1. (14 methods across its docs)
- **llms.txt**: Yes. https://docs.couchbase.com/llms.txt
- **Agent payments**: None found. The checkout: Reached the card form and stopped: Pay-as-you-go; Checkout: a Stripe card form; An invisible bot check loaded at checkout; Wallets: Google Pay.
- **Docs**: No gaps in what we read. Covers getting started, auth, error codes and code examples. 11 of 16 documented ways to do the core work worked when tried. https://docs.couchbase.com/home/index.html (462 of 5,457 pages read)

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

## Tips for agent builders

- Sign up at cloud.couchbase.com/sign-up with email and password, then make one Management API key in Settings > API Keys; everything after that works by API at https://cloudapi.cloud.couchbase.com/v4 with the key's token as Bearer.
- Use the freeTier paths for buckets on a free cluster (POST .../buckets/freeTier); the general bucket path needs replicas: 0 and its delete answers with a misleading 403.
- Turn on the Data API (PUT .../dataAPI {enableDataApi:true}, ready in about 5 minutes) for plain REST document, query and search calls with database credentials.
- Run the official MCP server with CB_MCP_TRANSPORT=http and CB_MCP_READ_ONLY_MODE=false to get its write tools; it negotiates protocol 2025-11-25 through initialize.
- For cbbackupmgr against Capella, create the repository with --disable-analytics --disable-cluster-analytics --disable-bucket-query --disable-cluster-query.
- Get the cluster CA with GET .../clusters/{id}/certificates for REST calls on ports 18091/18093/18094.
- When rotating an API key, use the returned token (not secretKey) as the Bearer value.

## How Couchbase compares

In Databases: 1st of 9 Databases products our agent got into on its own.

- [Prisma Postgres](https://agentdrives.com/tools/prisma.md?via=md): L4, Gets in on its own
- [Upstash](https://agentdrives.com/tools/upstash.md?via=md): L4, Gets in on its own
- [Airtable](https://agentdrives.com/tools/airtable.md?via=md): L4, Gets in on its own
- [Convex](https://agentdrives.com/tools/convex.md?via=md): L4, Gets in on its own
- [NocoDB](https://agentdrives.com/tools/nocodb.md?via=md): L4, Gets in on its own
- **Couchbase** (this page): L4, Gets in on its own
- [Firebase](https://agentdrives.com/tools/firebase-google.md?via=md): L3, Gets in on its own
- [Neon](https://agentdrives.com/tools/neon.md?via=md): L3, Gets in on its own

[All 15 in Databases](https://agentdrives.com/categories/databases?via=md)

Sorted by level, then work done, then time to the first call. [How we rank](https://agentdrives.com/method?via=md#order)

On our standard (18 criteria: against other databases products where we measured 5 or more, otherwise against every product we measured):

- [One way to page through lists](https://agentdrives.com/standard#api-03): **Ahead of most** 8 of 10 others (databases products) lower, 2 the same.
- [Every way in works for an agent](https://agentdrives.com/standard#in-11): **Ahead of most** 9 of 12 others (databases products) lower, 3 the same.
- [Export before leaving](https://agentdrives.com/standard#leave-03): **Ahead of most** 7 of 11 others (databases products) lower, 4 the same.
- [A request id on every answer](https://agentdrives.com/standard#err-07): **Ahead of most** 4 of 7 others (databases products) lower, 1 the same, 2 higher.
- [Versions an agent can pin](https://agentdrives.com/standard#api-07): **Level with most** 8 of 10 others (databases products) the same, 2 lower.
- [Fast to the first call](https://agentdrives.com/standard#in-03): **The same as all** 9 others (databases products).
- [Real status codes](https://agentdrives.com/standard#err-01): **The same as all** 11 others (databases products).
- [Errors say what to do next](https://agentdrives.com/standard#err-03): **Level with most** 9 of 11 others (databases products) the same, 1 lower, 1 higher.
- [WebMCP tools on the web app](https://agentdrives.com/standard#proto-02): **The same as all** 7 others (databases products).
- [Pages fit one fetch](https://agentdrives.com/standard#docs-08): **Level with most** 9 of 11 others (databases products) the same, 2 higher.
- [Docs an agent can navigate](https://agentdrives.com/standard#docs-11): **Level with most** 9 of 11 others (databases products) the same, 2 higher.
- [Limits known before they hit](https://agentdrives.com/standard#lim-02): **Level with most** 3 of 4 others (databases products) the same, 1 higher.
- [A path written for agents](https://agentdrives.com/standard#docs-10): **Level with most** 6 of 11 others (databases products) the same, 1 lower, 4 higher.
- [Errors a program can read](https://agentdrives.com/standard#err-02): **Level with most** 6 of 11 others (databases products) the same, 1 lower, 4 higher.
- [All validation problems at once](https://agentdrives.com/standard#err-04): **Level with most** 23 of 35 others (every product) the same, 12 higher.
- [Spend an agent can see and cap](https://agentdrives.com/standard#pay-05): **Behind most** 7 of 12 others (databases products) higher, 5 the same.
- [Answers sized for an agent](https://agentdrives.com/standard#api-04): **Behind most** 6 of 9 others (databases products) higher, 3 the same.
- [The best way in is built for agents](https://agentdrives.com/standard#in-12): **Behind most** 10 of 12 others (databases products) higher, 1 the same, 1 lower.

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

## For the team behind Couchbase

**9 things to fix for agents.**

Agent blockers: 2 found · Agent surfaces: 27 issues found.

Our agent got through Pay only in part. Prove Couchbase 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: Partly
- Leave: Clear

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

## Where Couchbase fits

- Category: [Data and storage](https://agentdrives.com/categories/data?via=md), [Databases](https://agentdrives.com/categories/databases?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), [Create or update a record](https://agentdrives.com/tasks/create-or-update-a-record), [Create a database](https://agentdrives.com/tasks/create-a-database), [Read usage and quotas](https://agentdrives.com/tasks/read-usage-and-quotas), [Delete a record](https://agentdrives.com/tasks/delete-a-record)
- Wall: [Sign-up or key only in a browser](https://agentdrives.com/access/get-in/sign-up) (our agent got in another way)
- 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)
- Rivals: [Airtable](https://agentdrives.com/tools/airtable?via=md) (L4), [Convex](https://agentdrives.com/tools/convex?via=md) (L4), [NocoDB](https://agentdrives.com/tools/nocodb?via=md) (L4)

## Cited in 1 article

- [CLI for AI agents: 11 of 14 databases have one](https://agentdrives.com/research/which-kinds-of-software-give-ai-agents-a-cli-or-mcp-server?via=md)

Data: [JSON](https://agentdrives.com/api/tools/couchbase.json?via=md) · [Markdown](https://agentdrives.com/tools/couchbase.md?via=md) · [History](https://agentdrives.com/tools/couchbase/history?via=md) · [Atom](https://agentdrives.com/feeds/tools/couchbase.atom) · [JSON Feed](https://agentdrives.com/feeds/tools/couchbase.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/couchbase?via=md#correct) or read [how corrections work](https://agentdrives.com/corrections?via=md).