--- title: "Fireworks AI MCP server for AI agents - actionlake" description: "Give Claude 112 Fireworks AI actions through one MCP server: evaluators, models, datasets. Your agent never holds the credential. 100,000 calls a month free." canonical: https://actionlake.com/connectors/fireworks-ai updated: 2026-10-23 --- Connectors · AI # Fireworks AI MCP server for AI agents. actionlake is an MCP gateway: one remote MCP server that connects Claude and other MCP clients to Fireworks AI and the rest of your systems. It gives your agent 112 Fireworks AI actions (51 reads, 45 writes and 16 that wait for a confirmation) and holds the Fireworks AI credential itself, so the agent never does. It paces Fireworks AI's limits and records every call. [Connect Fireworks AI 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: "fireworks_ai.accounts.list", intent: "check which Fireworks AI account is connected" } # List accounts · read · 1 upstream call ``` - 112 actions in one signed file - 51 read look, never change - 45 write create and update - 16 wait for a yes destructive, confirmed first Set up ## Connect Fireworks AI 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 Fireworks AI. 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. - `api_key` Fireworks API key (fw_...): app.fireworks.ai > Settings > API Keys ``` account { op: "connections.requirements", service: "fireworks_ai" } # 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: "fireworks_ai.accounts.list" } # list accounts · read · 1 upstream call check { action: "fireworks_ai.datasets.create", input: { … } } # free: inputs, permissions, cost and confirmations, nothing sent ``` Fireworks AI 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** `api_key` **Limits we pace to** 600 calls a minute, per connection **API covered** Fireworks Gateway REST API 5.10.0`api.fireworks.ai` **Built from** Fireworks AI's documentation, read 23 October 2026[docs.fireworks.ai](https://docs.fireworks.ai/llms.txt) [docs.fireworks.ai](https://docs.fireworks.ai/api-reference/introduction) **Service file** Version 1.0.0, signed (Ed25519) **Cached reads** 41 of 51 reads, for up to 5 minutes; a write clears what it affects **Shared entities** 5, so ask can line Fireworks AI up with your other systems **Data** Processed and stored in Sydney, Australia. No US region yet Capabilities ## What an AI agent can do in Fireworks AI. 112 actions in 32 areas. Every one carries its effect, so a key can allow the reads and hold the rest. - **51** read look, never change - **45** write create and update - **16** destructive wait for a confirmation 1. ### Evaluators 9 4 read · 4 write · 1 destructive List evaluators · Get a signed URL for an evaluator's build log · Get an evaluator · Get signed URLs for an evaluator's source code 2. ### Models 9 4 read · 4 write · 1 destructive Get a model · Create a model · Delete a model · Update a model 3. ### Datasets 8 3 read · 4 write · 1 destructive List datasets · Get a dataset · Create a dataset · Update a dataset 4. ### Deployments 7 2 read · 4 write · 1 destructive List deployments · Get a deployment · Create a deployment · Update a deployment 5. ### Dpo jobs 6 3 read · 2 write · 1 destructive List DPO jobs · Get a DPO job · Get a signed URL for a DPO job's metrics file · Create a DPO job 6. ### RFT jobs 6 2 read · 2 write · 2 destructive List reinforcement fine-tuning jobs · Get a reinforcement fine-tuning job · Create a reinforcement fine-tuning job · Resume a reinforcement fine-tuning job 7. ### Deployed models 5 2 read · 2 write · 1 destructive List LoRAs loaded on deployments · Get a loaded LoRA · Load a LoRA onto a deployment · Update a loaded LoRA 8. ### Evaluation jobs 5 3 read · 1 write · 1 destructive List evaluation jobs · Get a signed URL for an evaluation job's execution log · Get an evaluation job · Create an evaluation job **24 more areas · 57 actions** - Rlor trainer jobs **5** - Routers **5** - Secrets **5** - SFT jobs **5** - Batch inference jobs **4** - Responses **4** - Users **4** - API keys **3** - Deployment shape versions **3** - Quotas **3** - Accounts **2** - Policy settings **2** - Audit logs **1** - Billing summary **1** - Billing usage **1** - Chat completions **1** - Completions **1** - Decisions **1** - Deployment shapes **1** - Embeddings **1** - Messages **1** - Rerank **1** - Systemone **1** - Usage costs **1** Top actions ## The Fireworks AI API, shaped for agents. The ids your agent types, what each does, and the one name it shares with other systems. 38 of Fireworks AI's 38 action names are shared with at least one other connector, so an agent learns them once. | Action | What it does | Effect | Same name in | | --- | --- | --- | --- | | `fireworks_ai.users.list` List users | List Users. | read | `identity.user.list` 474 connectors | | `fireworks_ai.datasets.list` List datasets | List Datasets. | read | `analytics.dataset.list` 28 connectors | | `fireworks_ai.deployments.list` List deployments | List Deployments. | read | `infra.deployment.list` 8 connectors | | `fireworks_ai.users.get` Get a user | Get User. | read | `identity.user.get` 399 connectors | | `fireworks_ai.datasets.get` Get a dataset | Get Dataset. | read | `analytics.dataset.get` 22 connectors | | `fireworks_ai.datasets.create` Create a dataset | Create Dataset. | write | `analytics.dataset.create` 21 connectors | | `fireworks_ai.datasets.delete` Delete a dataset | Delete Dataset. | destructive | `analytics.dataset.delete` 23 connectors | | `fireworks_ai.deployments.get` Get a deployment | Get Deployment. | read | `infra.deployment.get` 8 connectors | | `fireworks_ai.api_keys.list` List API keys | List API Keys. | read | `identity.api_key.list` 96 connectors | | `fireworks_ai.accounts.list` List accounts | List Accounts. | read | `platform.tenant.list` 70 connectors | | `fireworks_ai.rerank.create` Rerank documents for a query | Rerank documents. | write | `analytics.inference.create` 17 connectors | | `fireworks_ai.messages.create` Create a message (Anthropic-compatible) | Create a Message. | write | `analytics.inference.create` 17 connectors | Try it ## Prompts to paste into Claude. Each one names the Fireworks AI actions it will call, so you know what it reads and what it changes before you send it. 1. Read read Who has access in Fireworks AI? List every user and their role. calls `fireworks_ai.users.list` 2. Write write Create a dataset in Fireworks AI. Show me exactly what you will send before you save it. calls `fireworks_ai.datasets.create` Governance for Fireworks AI ## Destructive actions wait for a yes. 16 Fireworks AI actions are destructive. These are all of them: - Delete a dataset `fireworks_ai.datasets.delete` - Delete a deployment `fireworks_ai.deployments.delete` - Delete an API key `fireworks_ai.api_keys.delete` - Delete a model `fireworks_ai.models.delete` - Delete a secret `fireworks_ai.secrets.delete` - Delete a batch inference job `fireworks_ai.batch_inference_jobs.delete` - Delete a router `fireworks_ai.routers.delete` - Delete a DPO job `fireworks_ai.dpo_jobs.delete` - Delete a reinforcement fine-tuning job `fireworks_ai.rft_jobs.delete` - Delete a supervised fine-tuning job `fireworks_ai.sft_jobs.delete` - Delete a stored response `fireworks_ai.responses.delete` - Delete an evaluator `fireworks_ai.evaluators.delete` - Unload a LoRA `fireworks_ai.deployed_models.delete` - Delete an evaluation job `fireworks_ai.evaluation_jobs.delete` - Delete a reinforcement fine-tuning step `fireworks_ai.rlor_trainer_jobs.delete` - Cancel a reinforcement fine-tuning job `fireworks_ai.rft_jobs.cancel` 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 fireworks_ai.datasets.delete confirm_policy: **human** Delete a dataset? **intent:** the user asked me to delete a dataset run suspended · waiting for a named person · key _fireworks-ai-agent_ cannot confirm itself In the console, or from an emailed signed link. a key for this connector ``` account { op: "keys.create", name: "fireworks-ai-agent", allowed_actions: ["fireworks_ai.*"], cap_calls: 2000, confirm_policy: "human_for_writes", approval_rules: [ { match: "fireworks_ai.datasets.delete", approver: "lead@example.com" }, { match: "fireworks_ai.deployments.delete", approver: "lead@example.com" }, { match: "fireworks_ai.api_keys.delete", approver: "lead@example.com" } ] } ``` Illustrative. Scoped to Fireworks AI only, capped at 2,000 upstream calls, every write confirmed, and the destructive ones approved by a named person. Cost ## What Fireworks AI costs on actionlake. Every Fireworks AI action is one upstream call, the only thing actionlake bills. Searching, `check`, `cost`, `account` and any call Fireworks AI 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 112 Fireworks AI actions - 112 actions page each page of a long list is one call - 41 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 Fireworks AI action. 112 actions by area: the id your agent calls, what it does, and its effect. **Evaluators 9** - `fireworks_ai.evaluators.build_log_endpoint.get` Get a signed URL for an evaluator's build log read - `fireworks_ai.evaluators.create` Create an evaluator write - `fireworks_ai.evaluators.delete` Delete an evaluator destructive - `fireworks_ai.evaluators.get` Get an evaluator read - `fireworks_ai.evaluators.list` List evaluators read - `fireworks_ai.evaluators.source_code_endpoint.get` Get signed URLs for an evaluator's source code read - `fireworks_ai.evaluators.update` Update an evaluator write - `fireworks_ai.evaluators.upload_endpoint.create` Get signed upload URLs for evaluator source code write - `fireworks_ai.evaluators.validate_upload` Validate an uploaded evaluator write **Models 9** - `fireworks_ai.models.create` Create a model write - `fireworks_ai.models.delete` Delete a model destructive - `fireworks_ai.models.download_endpoint.get` Get signed download URLs for model files read - `fireworks_ai.models.get` Get a model read - `fireworks_ai.models.list` List models read - `fireworks_ai.models.prepare` Prepare a model for a precision write - `fireworks_ai.models.update` Update a model write - `fireworks_ai.models.upload_endpoint.create` Get signed upload URLs for model files write - `fireworks_ai.models.validate_upload` Validate an uploaded model read **Datasets 8** - `fireworks_ai.datasets.create` Create a dataset write - `fireworks_ai.datasets.delete` Delete a dataset destructive - `fireworks_ai.datasets.download_endpoint.get` Get signed download URLs for dataset files read - `fireworks_ai.datasets.get` Get a dataset read - `fireworks_ai.datasets.list` List datasets read - `fireworks_ai.datasets.update` Update a dataset write - `fireworks_ai.datasets.upload_endpoint.create` Get signed upload URLs for dataset files write - `fireworks_ai.datasets.validate_upload` Validate an uploaded dataset write **Deployments 7** - `fireworks_ai.deployments.create` Create a deployment write - `fireworks_ai.deployments.delete` Delete a deployment destructive - `fireworks_ai.deployments.get` Get a deployment read - `fireworks_ai.deployments.list` List deployments read - `fireworks_ai.deployments.scale` Scale a deployment write - `fireworks_ai.deployments.undelete` Undelete a deployment write - `fireworks_ai.deployments.update` Update a deployment write **Dpo jobs 6** - `fireworks_ai.dpo_jobs.create` Create a DPO job write - `fireworks_ai.dpo_jobs.delete` Delete a DPO job destructive - `fireworks_ai.dpo_jobs.get` Get a DPO job read - `fireworks_ai.dpo_jobs.list` List DPO jobs read - `fireworks_ai.dpo_jobs.metrics_file_endpoint.get` Get a signed URL for a DPO job's metrics file read - `fireworks_ai.dpo_jobs.resume` Resume a DPO job write **RFT jobs 6** - `fireworks_ai.rft_jobs.cancel` Cancel a reinforcement fine-tuning job destructive - `fireworks_ai.rft_jobs.create` Create a reinforcement fine-tuning job write - `fireworks_ai.rft_jobs.delete` Delete a reinforcement fine-tuning job destructive - `fireworks_ai.rft_jobs.get` Get a reinforcement fine-tuning job read - `fireworks_ai.rft_jobs.list` List reinforcement fine-tuning jobs read - `fireworks_ai.rft_jobs.resume` Resume a reinforcement fine-tuning job write **Deployed models 5** - `fireworks_ai.deployed_models.create` Load a LoRA onto a deployment write - `fireworks_ai.deployed_models.delete` Unload a LoRA destructive - `fireworks_ai.deployed_models.get` Get a loaded LoRA read - `fireworks_ai.deployed_models.list` List LoRAs loaded on deployments read - `fireworks_ai.deployed_models.update` Update a loaded LoRA write **Evaluation jobs 5** - `fireworks_ai.evaluation_jobs.create` Create an evaluation job write - `fireworks_ai.evaluation_jobs.delete` Delete an evaluation job destructive - `fireworks_ai.evaluation_jobs.execution_log_endpoint.get` Get a signed URL for an evaluation job's execution log read - `fireworks_ai.evaluation_jobs.get` Get an evaluation job read - `fireworks_ai.evaluation_jobs.list` List evaluation jobs read **Rlor trainer jobs 5** - `fireworks_ai.rlor_trainer_jobs.create` Create a reinforcement fine-tuning step write - `fireworks_ai.rlor_trainer_jobs.delete` Delete a reinforcement fine-tuning step destructive - `fireworks_ai.rlor_trainer_jobs.get` Get a reinforcement fine-tuning step read - `fireworks_ai.rlor_trainer_jobs.list` List reinforcement fine-tuning steps read - `fireworks_ai.rlor_trainer_jobs.resume` Resume a reinforcement fine-tuning step write **Routers 5** - `fireworks_ai.routers.create` Create a router write - `fireworks_ai.routers.delete` Delete a router destructive - `fireworks_ai.routers.get` Get a router read - `fireworks_ai.routers.list` List routers read - `fireworks_ai.routers.update` Update a router write **Secrets 5** - `fireworks_ai.secrets.create` Create a secret write - `fireworks_ai.secrets.delete` Delete a secret destructive - `fireworks_ai.secrets.get` Get a secret read - `fireworks_ai.secrets.list` List secrets read - `fireworks_ai.secrets.update` Update a secret write **SFT jobs 5** - `fireworks_ai.sft_jobs.create` Create a supervised fine-tuning job write - `fireworks_ai.sft_jobs.delete` Delete a supervised fine-tuning job destructive - `fireworks_ai.sft_jobs.get` Get a supervised fine-tuning job read - `fireworks_ai.sft_jobs.list` List supervised fine-tuning jobs read - `fireworks_ai.sft_jobs.resume` Resume a supervised fine-tuning job write **Batch inference jobs 4** - `fireworks_ai.batch_inference_jobs.create` Create a batch inference job write - `fireworks_ai.batch_inference_jobs.delete` Delete a batch inference job destructive - `fireworks_ai.batch_inference_jobs.get` Get a batch inference job read - `fireworks_ai.batch_inference_jobs.list` List batch inference jobs read **Responses 4** - `fireworks_ai.responses.create` Create a model response write - `fireworks_ai.responses.delete` Delete a stored response destructive - `fireworks_ai.responses.get` Get a stored response read - `fireworks_ai.responses.list` List stored responses read **Users 4** - `fireworks_ai.users.create` Create a user or service account write - `fireworks_ai.users.get` Get a user read - `fireworks_ai.users.list` List users read - `fireworks_ai.users.update` Update a user write **API keys 3** - `fireworks_ai.api_keys.create` Create an API key write - `fireworks_ai.api_keys.delete` Delete an API key destructive - `fireworks_ai.api_keys.list` List API keys read **Deployment shape versions 3** - `fireworks_ai.deployment_shape_versions.get` Get a deployment shape version read - `fireworks_ai.deployment_shape_versions.list` List deployment shape versions read - `fireworks_ai.deployment_shape_versions.match` Match deployment shape versions to a deployment request read **Quotas 3** - `fireworks_ai.quotas.get` Get a quota read - `fireworks_ai.quotas.list` List quotas read - `fireworks_ai.quotas.update` Update a quota write **Accounts 2** - `fireworks_ai.accounts.get` Get an account read - `fireworks_ai.accounts.list` List accounts read **Policy settings 2** - `fireworks_ai.policy_settings.get` Get policy settings read - `fireworks_ai.policy_settings.update` Update policy settings write **Audit logs 1** - `fireworks_ai.audit_logs.list` List user audit logs read **Billing summary 1** - `fireworks_ai.billing_summary.get` Get billing summary read **Billing usage 1** - `fireworks_ai.billing_usage.get` Get account usage read **Chat completions 1** - `fireworks_ai.chat_completions.create` Create a chat completion write **Completions 1** - `fireworks_ai.completions.create` Create a text completion write **Decisions 1** - `fireworks_ai.decisions.create` Answer typed questions (Decisions API) write **Deployment shapes 1** - `fireworks_ai.deployment_shapes.get` Get a deployment shape read **Embeddings 1** - `fireworks_ai.embeddings.create` Create embeddings write **Messages 1** - `fireworks_ai.messages.create` Create a message (Anthropic-compatible) write **Rerank 1** - `fireworks_ai.rerank.create` Rerank documents for a query write **Systemone 1** - `fireworks_ai.systemone.create` Answer typed questions (System One API) write **Usage costs 1** - `fireworks_ai.usage_costs.query` Query grouped usage cost subtotals read Questions ## The Fireworks AI MCP server, answered. **Is there a Fireworks AI MCP server for Claude?** Yes. actionlake is an MCP gateway: a remote MCP server that gives Claude, or another MCP client, 112 Fireworks AI actions across evaluators, models, datasets and deployments, through the same eight tools it uses for every other connector. actionlake is independent; it is not made or endorsed by Fireworks AI. **What do I need to connect Fireworks AI?** Fireworks API key (fw_...): app.fireworks.ai > Settings > API Keys. The free call account { op: "connections.requirements", service: "fireworks_ai" } lists exactly what to paste. **Can an AI agent delete or change things in Fireworks AI?** Only if its key allows it, and only with an explicit confirmation. 16 Fireworks AI actions are destructive, including delete a dataset, delete a deployment and delete an API key. 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 Fireworks AI API limit?** actionlake paces every call to the limits in the Fireworks AI service file: 600 calls a minute per connection. Repeated reads are answered from cache (41 Fireworks AI actions can be), which never reaches Fireworks AI. **What does a Fireworks AI action cost?** One upstream call, the only unit actionlake bills. A list that pages bills one call per page (112 Fireworks AI actions page). Cache hits, searches, dry runs and calls Fireworks AI 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 Fireworks AI API does it cover?** The Fireworks Gateway REST API 5.10.0 at api.fireworks.ai. It is built from Fireworks AI's documentation as read on 23 October 2026. 112 actions in 32 areas, each one in the signed service file (version 1.0.0). **Does actionlake keep my Fireworks AI 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 Fireworks AI 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 Fireworks AI Shares the most action names with - [OpenRouter 16](https://actionlake.com/connectors/openrouter) - [Together AI 14](https://actionlake.com/connectors/together-ai) - [Mistral AI 13](https://actionlake.com/connectors/mistral-ai) - [Hugging Face 13](https://actionlake.com/connectors/hugging-face) - [Google Gemini API (AI Studio) 10](https://actionlake.com/connectors/google-gemini-api-ai-studio) - [Langfuse 9](https://actionlake.com/connectors/langfuse) Joins with, on shared entities - [Grafana 3](https://actionlake.com/connectors/grafana) - [Axiom 3](https://actionlake.com/connectors/axiom) - [Slack 2](https://actionlake.com/connectors/slack) - [Gusto 2](https://actionlake.com/connectors/gusto) - [Microsoft 365 Users 2](https://actionlake.com/connectors/microsoft-365-users) - [Cloudflare 2](https://actionlake.com/connectors/cloudflare) 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 Fireworks AI; the name is the vendor's own. Counts are from the Fireworks AI service file, version 1.0.0. ## Give your agent Fireworks AI today. [Connect Fireworks AI 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/fireworks-ai More: https://actionlake.com/llms.txt