--- title: "Braintrust MCP server for AI agents - actionlake" description: "Give Claude 167 Braintrust actions through one MCP server: dataset, experiment, acl. Your agent never holds the credential. 100,000 calls a month free." canonical: https://actionlake.com/connectors/braintrust updated: 2026-10-23 --- Connectors · AI # Braintrust MCP server for AI agents. actionlake is an MCP gateway: one remote MCP server that connects Claude and other MCP clients to Braintrust and the rest of your systems. It gives your agent 167 Braintrust actions (62 reads, 78 writes and 27 that wait for a confirmation) and holds the Braintrust credential itself, so the agent never does. It paces Braintrust's limits and records every call. [Connect Braintrust free](https://actionlake.com/console?start=sandbox) [See pricing](https://actionlake.com/pricing) 100,000 free calls a month. No card, no email, no password: press Start a free sandbox on the next screen and keep the key it shows you. `claude mcp add --transport http actionlake https://connect.actionlake.com/mcp` the first call ``` do { action: "braintrust.user.list", intent: "check which Braintrust account is connected" } # List users · read · 1 upstream call ``` - 167 actions in one signed file - 62 read look, never change - 78 write create and update - 27 wait for a yes destructive, confirmed first Set up ## Connect Braintrust in three steps. 1. 01 ### Add actionlake to Claude. Run the command once in Claude Code; its consent page can create a free sandbox: 100,000 calls a month, no card. In the Claude app, add a custom connector with the URL below and paste your key on the consent page. Creating a new sandbox from the Claude app is not yet tested by us. `https://connect.actionlake.com/mcp` 2. 02 ### Connect Braintrust. Bearer token. No consent screen: your agent can connect it for you, but the key then passes through your chat. To keep it out of the model's context, paste it in the console instead. actionlake stores what you paste write-only, sealed under your tenant's key. - `access_token` Braintrust API key (sk-...) ``` account { op: "connections.requirements", service: "braintrust" } # free: exactly what to paste, and what is found for you ``` 3. 03 ### Make the first call. Check before a write. Ask in plain words; Claude picks the action. Before anything changes, `check` validates the call and prices it, for free. ``` do { action: "braintrust.user.list" } # list users · read · 1 upstream call check { action: "braintrust.acl.create", input: { … } } # free: inputs, permissions, cost and confirmations, nothing sent ``` Braintrust at a glance **Sign-in** Bearer token. No consent screen: your agent can connect it for you, but the key then passes through your chat. To keep it out of the model's context, paste it in the console instead. **You provide** `access_token` **Limits we pace to** 300 calls a minute, per connection **API covered** Braintrust API 1.0.0 (OpenAPI 3.0.3)`api.braintrust.dev` **Built from** Braintrust's documentation, read 23 October 2026[OpenAPI spec (GitHub)](https://raw.githubusercontent.com/braintrustdata/braintrust-openapi/main/openapi/spec.json) [braintrust.dev](https://www.braintrust.dev/docs/api-reference) **Service file** Version 1.0.0, signed (Ed25519) **Cached reads** 47 of 62 reads, for up to 5 minutes; a write clears what it affects **Shared entities** 5, so ask can line Braintrust up with your other systems **Data** Processed and stored in Sydney, Australia. No US region yet Capabilities ## What an AI agent can do in Braintrust. 167 actions in 30 areas. Every one carries its effect, so a key can allow the reads and hold the rest. - **62** read look, never change - **78** write create and update - **27** destructive wait for a confirmation 1. ### Dataset 10 5 read · 4 write · 1 destructive List datasets · Get dataset · Fetch dataset (POST form) · Fetch dataset (GET form) 2. ### Experiment 10 5 read · 4 write · 1 destructive List experiments · Get experiment · Fetch experiment (POST form) · Fetch experiment (GET form) 3. ### Acl 7 3 read · 2 write · 2 destructive List acls · List org acls · Get acl · Create acl 4. ### AI secret 7 2 read · 3 write · 2 destructive List ai_secrets · Get ai_secret · Create ai_secret · Partially update ai_secret 5. ### Function 7 2 read · 4 write · 1 destructive List functions · Get function · Create function · Invoke function 6. ### Agent 6 2 read · 3 write · 1 destructive List agents · Get agent · Create agent · Partially update agent 7. ### Dataset snapshot 6 2 read · 3 write · 1 destructive List dataset_snapshots · Get dataset_snapshot · Create dataset_snapshot · Partially update dataset_snapshot 8. ### Env var 6 2 read · 3 write · 1 destructive List env_vars · Get env_var · Create env_var · Partially update env_var **22 more areas · 108 actions** - Group **6** - MCP server **6** - Org automation **6** - Project automation **6** - Project group **6** - Project score **6** - Project tag **6** - Prompt **6** - Role **6** - Service token **6** - Span iframe **6** - View **6** - Environment **5** - Organization **5** - Project **5** - Proxy **5** - Simulated user **5** - Project logs **4** - API key **3** - User **2** - Cross object **1** - Eval **1** Top actions ## The Braintrust API, shaped for agents. The ids your agent types, what each does, and the one name it shares with other systems. 16 of Braintrust's 16 action names are shared with at least one other connector, so an agent learns them once. | Action | What it does | Effect | Same name in | | --- | --- | --- | --- | | `braintrust.user.list` List users | List out all users. | read | `identity.user.list` 474 connectors | | `braintrust.group.list` List groups | List out all groups. | read | `identity.group.list` 262 connectors | | `braintrust.user.get` Get user | Get a user object by its id. | read | `identity.user.get` 399 connectors | | `braintrust.group.get` Get group | Get a group object by its id. | read | `identity.group.get` 212 connectors | | `braintrust.dataset.list` List datasets | List out all datasets. | read | `analytics.dataset.list` 28 connectors | | `braintrust.project.list` List projects | List out all projects. | read | `work.project.list` 109 connectors | | `braintrust.role.list` List roles | List out all roles. | read | `identity.role.list` 184 connectors | | `braintrust.api_key.list` List api_keys | List out all API keys. | read | `identity.api_key.list` 96 connectors | | `braintrust.acl.create` Create acl | Create a new acl. | write | this connector | | `braintrust.proxy.auto` Proxy a model to chat/completions or completions automatically | Proxy a request to either chat/completions or completions automatically based on the model. | write | this connector | | `braintrust.eval.launch` Launch an eval | Launch an evaluation. | write | this connector | | `braintrust.acl.delete` Delete acl | Delete an acl object by its id. | destructive | this connector | Try it ## Prompts to paste into Claude. Each one names the Braintrust actions it will call, so you know what it reads and what it changes before you send it. 1. Read read Give me a one-line status for every active Braintrust project. calls `braintrust.project.list` 2. Write write Create acl in Braintrust. Show me exactly what you will send before you save it. calls `braintrust.acl.create` Governance for Braintrust ## Destructive actions wait for a yes. 27 Braintrust actions are destructive. These are all of them: - Delete acl `braintrust.acl.delete` - Delete role `braintrust.role.delete` - Delete view `braintrust.view.delete` - Delete agent `braintrust.agent.delete` - Delete group `braintrust.group.delete` - Delete prompt `braintrust.prompt.delete` - Delete api_key `braintrust.api_key.delete` - Delete dataset `braintrust.dataset.delete` - Delete env_var `braintrust.env_var.delete` - Delete project `braintrust.project.delete` - Delete function `braintrust.function.delete` - Delete ai_secret `braintrust.ai_secret.delete` - Delete experiment `braintrust.experiment.delete` - Delete mcp_server `braintrust.mcp_server.delete` - Delete environment `braintrust.environment.delete` - Delete project_tag `braintrust.project_tag.delete` - Delete span_iframe `braintrust.span_iframe.delete` - Delete single acl `braintrust.acl.delete_matching` - Delete project_group `braintrust.project_group.delete` - Delete project_score `braintrust.project_score.delete` - Delete service_token `braintrust.service_token.delete` - Delete org_automation `braintrust.org_automation.delete` - Delete simulated user `braintrust.simulated_user.delete` - Delete dataset_snapshot `braintrust.dataset_snapshot.delete` - Delete single ai_secret `braintrust.ai_secret.delete_by_name` - Delete project_automation `braintrust.project_automation.delete` - Delete single service_token `braintrust.service_token.delete_matching` A destructive action runs only with `confirm: true`. Under the `human` policy the agent cannot confirm for itself: the run stops until a named person approves, in the console or from a signed link in an email. The cap on a key is refused before the call is sent, never after. do braintrust.acl.delete confirm_policy: **human** Delete acl? **intent:** the user asked me to delete acl run suspended · waiting for a named person · key _braintrust-agent_ cannot confirm itself In the console, or from an emailed signed link. a key for this connector ``` account { op: "keys.create", name: "braintrust-agent", allowed_actions: ["braintrust.*"], cap_calls: 2000, confirm_policy: "human_for_writes", approval_rules: [ { match: "braintrust.acl.delete", approver: "lead@example.com" }, { match: "braintrust.role.delete", approver: "lead@example.com" }, { match: "braintrust.view.delete", approver: "lead@example.com" } ] } ``` Illustrative. Scoped to Braintrust only, capped at 2,000 upstream calls, every write confirmed, and the destructive ones approved by a named person. Cost ## What Braintrust costs on actionlake. Every Braintrust action is one upstream call, the only thing actionlake bills. Searching, `check`, `cost`, `account` and any call Braintrust refused with 401, 403, 429 or a 5xx are free. Paid plans start at US$49 a month for 250,000 calls, with overage from US$0.30 down to US$0.18 per 1,000. - 1 upstream call per action all 167 Braintrust actions - 167 actions page each page of a long list is one call - 47 reads can be cached a cache hit costs nothing - 100,000 calls a month, free paid plans from US$49 a month for 250,000 [See pricing](https://actionlake.com/pricing) Action index ## Every Braintrust action. 167 actions by area: the id your agent calls, what it does, and its effect. **Dataset 10** - `braintrust.dataset.create` Create dataset write - `braintrust.dataset.delete` Delete dataset destructive - `braintrust.dataset.feedback` Feedback for dataset events write - `braintrust.dataset.fetch` Fetch dataset (POST form) read - `braintrust.dataset.fetch_get` Fetch dataset (GET form) read - `braintrust.dataset.get` Get dataset read - `braintrust.dataset.insert` Insert dataset events write - `braintrust.dataset.list` List datasets read - `braintrust.dataset.summarize` Summarize dataset read - `braintrust.dataset.update` Partially update dataset write **Experiment 10** - `braintrust.experiment.create` Create experiment write - `braintrust.experiment.delete` Delete experiment destructive - `braintrust.experiment.feedback` Feedback for experiment events write - `braintrust.experiment.fetch` Fetch experiment (POST form) read - `braintrust.experiment.fetch_get` Fetch experiment (GET form) read - `braintrust.experiment.get` Get experiment read - `braintrust.experiment.insert` Insert experiment events write - `braintrust.experiment.list` List experiments read - `braintrust.experiment.summarize` Summarize experiment read - `braintrust.experiment.update` Partially update experiment write **Acl 7** - `braintrust.acl.batch_update` Batch update acls write - `braintrust.acl.create` Create acl write - `braintrust.acl.delete` Delete acl destructive - `braintrust.acl.delete_matching` Delete single acl destructive - `braintrust.acl.get` Get acl read - `braintrust.acl.list` List acls read - `braintrust.acl.list_org` List org acls read **AI secret 7** - `braintrust.ai_secret.create` Create ai_secret write - `braintrust.ai_secret.delete` Delete ai_secret destructive - `braintrust.ai_secret.delete_by_name` Delete single ai_secret destructive - `braintrust.ai_secret.get` Get ai_secret read - `braintrust.ai_secret.list` List ai_secrets read - `braintrust.ai_secret.replace` Create or replace ai_secret write - `braintrust.ai_secret.update` Partially update ai_secret write **Function 7** - `braintrust.function.create` Create function write - `braintrust.function.delete` Delete function destructive - `braintrust.function.get` Get function read - `braintrust.function.invoke` Invoke function write - `braintrust.function.list` List functions read - `braintrust.function.replace` Create or replace function write - `braintrust.function.update` Partially update function write **Agent 6** - `braintrust.agent.create` Create agent write - `braintrust.agent.delete` Delete agent destructive - `braintrust.agent.get` Get agent read - `braintrust.agent.list` List agents read - `braintrust.agent.replace` Create or replace agent write - `braintrust.agent.update` Partially update agent write **Dataset snapshot 6** - `braintrust.dataset_snapshot.create` Create dataset_snapshot write - `braintrust.dataset_snapshot.delete` Delete dataset_snapshot destructive - `braintrust.dataset_snapshot.get` Get dataset_snapshot read - `braintrust.dataset_snapshot.list` List dataset_snapshots read - `braintrust.dataset_snapshot.replace` Create or replace dataset_snapshot write - `braintrust.dataset_snapshot.update` Partially update dataset_snapshot write **Env var 6** - `braintrust.env_var.create` Create env_var write - `braintrust.env_var.delete` Delete env_var destructive - `braintrust.env_var.get` Get env_var read - `braintrust.env_var.list` List env_vars read - `braintrust.env_var.replace` Create or replace env_var write - `braintrust.env_var.update` Partially update env_var write **Group 6** - `braintrust.group.create` Create group write - `braintrust.group.delete` Delete group destructive - `braintrust.group.get` Get group read - `braintrust.group.list` List groups read - `braintrust.group.replace` Create or replace group write - `braintrust.group.update` Partially update group write **MCP server 6** - `braintrust.mcp_server.create` Create mcp_server write - `braintrust.mcp_server.delete` Delete mcp_server destructive - `braintrust.mcp_server.get` Get mcp_server read - `braintrust.mcp_server.list` List mcp_servers read - `braintrust.mcp_server.replace` Create or replace mcp_server write - `braintrust.mcp_server.update` Partially update mcp_server write **Org automation 6** - `braintrust.org_automation.create` Create org_automation write - `braintrust.org_automation.delete` Delete org_automation destructive - `braintrust.org_automation.get` Get org_automation read - `braintrust.org_automation.list` List org_automations read - `braintrust.org_automation.replace` Create or replace org_automation write - `braintrust.org_automation.update` Partially update org_automation write **Project automation 6** - `braintrust.project_automation.create` Create project_automation write - `braintrust.project_automation.delete` Delete project_automation destructive - `braintrust.project_automation.get` Get project_automation read - `braintrust.project_automation.list` List project_automations read - `braintrust.project_automation.replace` Create or replace project_automation write - `braintrust.project_automation.update` Partially update project_automation write **Project group 6** - `braintrust.project_group.create` Create project_group write - `braintrust.project_group.delete` Delete project_group destructive - `braintrust.project_group.get` Get project_group read - `braintrust.project_group.list` List project_groups read - `braintrust.project_group.replace` Create or replace project_group write - `braintrust.project_group.update` Partially update project_group write **Project score 6** - `braintrust.project_score.create` Create project_score write - `braintrust.project_score.delete` Delete project_score destructive - `braintrust.project_score.get` Get project_score read - `braintrust.project_score.list` List project_scores read - `braintrust.project_score.replace` Create or replace project_score write - `braintrust.project_score.update` Partially update project_score write **Project tag 6** - `braintrust.project_tag.create` Create project_tag write - `braintrust.project_tag.delete` Delete project_tag destructive - `braintrust.project_tag.get` Get project_tag read - `braintrust.project_tag.list` List project_tags read - `braintrust.project_tag.replace` Create or replace project_tag write - `braintrust.project_tag.update` Partially update project_tag write **Prompt 6** - `braintrust.prompt.create` Create prompt write - `braintrust.prompt.delete` Delete prompt destructive - `braintrust.prompt.get` Get prompt read - `braintrust.prompt.list` List prompts read - `braintrust.prompt.replace` Create or replace prompt write - `braintrust.prompt.update` Partially update prompt write **Role 6** - `braintrust.role.create` Create role write - `braintrust.role.delete` Delete role destructive - `braintrust.role.get` Get role read - `braintrust.role.list` List roles read - `braintrust.role.replace` Create or replace role write - `braintrust.role.update` Partially update role write **Service token 6** - `braintrust.service_token.create` Create service_token write - `braintrust.service_token.delete` Delete service_token destructive - `braintrust.service_token.delete_matching` Delete single service_token destructive - `braintrust.service_token.get` Get service_token read - `braintrust.service_token.list` List service_tokens read - `braintrust.service_token.replace` Create or replace service_token write **Span iframe 6** - `braintrust.span_iframe.create` Create span_iframe write - `braintrust.span_iframe.delete` Delete span_iframe destructive - `braintrust.span_iframe.get` Get span_iframe read - `braintrust.span_iframe.list` List span_iframes read - `braintrust.span_iframe.replace` Create or replace span_iframe write - `braintrust.span_iframe.update` Partially update span_iframe write **View 6** - `braintrust.view.create` Create view write - `braintrust.view.delete` Delete view destructive - `braintrust.view.get` Get view read - `braintrust.view.list` List views read - `braintrust.view.replace` Create or replace view write - `braintrust.view.update` Partially update view write **Environment 5** - `braintrust.environment.create` Create environment write - `braintrust.environment.delete` Delete environment destructive - `braintrust.environment.get` Get environment read - `braintrust.environment.list` List environments read - `braintrust.environment.update` Partially update environment write **Organization 5** - `braintrust.organization.get` Get organization read - `braintrust.organization.list` List organizations read - `braintrust.organization.list_api_keys` List organization API keys read - `braintrust.organization.update` Partially update organization write - `braintrust.organization.update_members` Modify organization membership write **Project 5** - `braintrust.project.create` Create project write - `braintrust.project.delete` Delete project destructive - `braintrust.project.get` Get project read - `braintrust.project.list` List projects read - `braintrust.project.update` Partially update project write **Proxy 5** - `braintrust.proxy.auto` Proxy a model to chat/completions or completions automatically write - `braintrust.proxy.chat_completions` Proxy chat/completions write - `braintrust.proxy.completions` Proxy completions write - `braintrust.proxy.create_credentials` Create temporary credential write - `braintrust.proxy.embeddings` Proxy embeddings write **Simulated user 5** - `braintrust.simulated_user.create` Create simulated user write - `braintrust.simulated_user.delete` Delete simulated user destructive - `braintrust.simulated_user.get` Get simulated user read - `braintrust.simulated_user.list` List simulated users read - `braintrust.simulated_user.update` Update simulated user write **Project logs 4** - `braintrust.project_logs.feedback` Feedback for project logs events write - `braintrust.project_logs.fetch` Fetch project logs (POST form) read - `braintrust.project_logs.fetch_get` Fetch project logs (GET form) read - `braintrust.project_logs.insert` Insert project logs events write **API key 3** - `braintrust.api_key.delete` Delete api_key destructive - `braintrust.api_key.get` Get api_key read - `braintrust.api_key.list` List api_keys read **User 2** - `braintrust.user.get` Get user read - `braintrust.user.list` List users read **Cross object 1** - `braintrust.cross_object.insert` Cross-object insert write **Eval 1** - `braintrust.eval.launch` Launch an eval write Questions ## The Braintrust MCP server, answered. **Is there a Braintrust MCP server for Claude?** Yes. actionlake is an MCP gateway: a remote MCP server that gives Claude, or another MCP client, 167 Braintrust actions across dataset, experiment, acl and AI secret, through the same eight tools it uses for every other connector. actionlake is independent; it is not made or endorsed by Braintrust. **What do I need to connect Braintrust?** Braintrust API key (sk-...). The free call account { op: "connections.requirements", service: "braintrust" } lists exactly what to paste. **Can an AI agent delete or change things in Braintrust?** Only if its key allows it, and only with an explicit confirmation. 27 Braintrust actions are destructive, including delete acl, delete role and delete view. Each runs only with confirm: true. Under the human confirm policy, or an approval rule on the key, the run waits until a named person approves it. **Will an AI agent use up my Braintrust API limit?** actionlake paces every call to the limits in the Braintrust service file: 300 calls a minute per connection. Repeated reads are answered from cache (47 Braintrust actions can be), which never reaches Braintrust. **What does a Braintrust action cost?** One upstream call, the only unit actionlake bills. A list that pages bills one call per page (167 Braintrust actions page). Cache hits, searches, dry runs and calls Braintrust refused with 401, 403, 429 or a 5xx are free. The Free plan includes 100,000 calls a month with no card. Paid plans start at US$49 a month for 250,000 calls, with overage from US$0.30 down to US$0.18 per 1,000. **Which Braintrust API does it cover?** The Braintrust API 1.0.0 (OpenAPI 3.0.3) at api.braintrust.dev. It is built from Braintrust's documentation as read on 23 October 2026. 167 actions in 30 areas, each one in the signed service file (version 1.0.0). **Does actionlake keep my Braintrust data, and where?** actionlake processes and stores data in Sydney, Australia; there is no US region yet. Reads may be cached for up to 5 minutes, and a write clears the reads it affects. Each run's request and response is kept, sealed, for your plan's trace window: 1 day on Free, 3 days on Starter, 7 days on Pro, 14 days on Team, 30 days on Business and 30 days on Enterprise. Your Braintrust credential is sealed under your tenant's key, never logged and never returned. Your traffic is not used to train models. **Which AI clients can use it?** Claude Code, tested by us end to end, and the Claude app as a custom connector, tested by us with an existing key. Other MCP clients that support remote servers over Streamable HTTP (MCP revisions 2025-06-18 and 2025-11-25) and OAuth dynamic client registration, or a bearer header, should connect the same way; we have not tested them. Or plain REST, under the same action names. ## More connectors like Braintrust Shares the most action names with - [Fireworks AI 7](https://actionlake.com/connectors/fireworks-ai) - [Langfuse 5](https://actionlake.com/connectors/langfuse) - [Magnific 4](https://actionlake.com/connectors/magnific) - [Weaviate Cloud 4](https://actionlake.com/connectors/weaviate-cloud) - [LlamaCloud 4](https://actionlake.com/connectors/llamacloud) - [Portkey 4](https://actionlake.com/connectors/portkey) Joins with, on shared entities - [Octopus Deploy 4](https://actionlake.com/connectors/octopus-deploy) - [Tableau 4](https://actionlake.com/connectors/tableau) - [Grafana 4](https://actionlake.com/connectors/grafana) - [Axiom 4](https://actionlake.com/connectors/axiom) - [Domo 4](https://actionlake.com/connectors/domo) - [HaloPSA 3](https://actionlake.com/connectors/halopsa) Browse - [All AI connectors](https://actionlake.com/connectors/category/ai) - [All connectors](https://actionlake.com/connectors) - [How actionlake compares](https://actionlake.com/compare) actionlake is independent. It is not made or endorsed by Braintrust; the name is the vendor's own. Counts are from the Braintrust service file, version 1.0.0. ## Give your agent Braintrust today. [Connect Braintrust free](https://actionlake.com/console?start=sandbox) [See pricing](https://actionlake.com/pricing) Start free - 100,000 calls/month, no card. No signup. Hard caps on every key. Cached answers are free. --- Source: https://actionlake.com/connectors/braintrust More: https://actionlake.com/llms.txt