Fireworks AI for AI agents
Fireworks AI provides serverless inference and fine-tuning for open models, used by developers building AI-powered features who need speed and control over model choice. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.
What agents do in Fireworks AI.
Run a chat completion
Send a prompt to a deployed model to generate a response for an application feature.
Check a deployment's status
Get a deployment's status to confirm a fine-tuned model is live before routing traffic to it.
Read-only model catalog checks
List available models as a read-only step before choosing one for a new use case.
9 Fireworks AI actions, graded by risk.
Every Fireworks AI action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.
read
5Look things up. Allowed by default.
- fireworks_ai.get_modelGet Model
- fireworks_ai.list_modelsList Models
- fireworks_ai.list_accountsList Accounts
- fireworks_ai.get_deploymentGet Deployment
- fireworks_ai.list_deploymentsList Deployments
write
4Create and change things. Allow, or ask the user first.
- fireworks_ai.create_embeddingsCreate Embeddings
- fireworks_ai.create_chat_completionCreate Chat Completion
- fireworks_ai.rerank_documentsRerank Documents
- fireworks_ai.generate_model_responseGenerate Model Response
destructive
0Delete, cancel or archive. Ask first, or deny outright.
- No destructive actions.
Fireworks AI in three steps.
- 01Your users connect Fireworks AIThey add their Fireworks AI api key on Arc0 Connect, under your brand. It goes straight into the vault.
- 02You set the rulesReads run, and writes like “create Embeddings” can wait for the user to approve.
- 03Any agent can actYour agent calls Fireworks AI through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
await arc0.policies.set('fireworks_ai', { read: 'allow', write: 'ask', // create_embeddings destructive: 'deny', }) # Claude Code: the same connection, one URL $ claude mcp add --transport http arc0 \ https://mcp.arc0.ai/u/u_8f2
How Fireworks AI connects.
Users add their Fireworks AI api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
The same Fireworks AI connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →
- AUTH
- API key
- CREDENTIALS
- Per-tenant encrypted vault
- MODEL SEES
- Results only, never credentials
- AUDIT LOG
- Every call, on every plan
Use Fireworks AI from any agent.
ClaudeChatGPTCursor
Codex
VS CodeOpenAI Agents SDKClaude Agent SDK
Vercel AI SDK
Mastra
LangGraph
Fireworks AI and Arc0, answered.
Can I use Fireworks AI with Claude, ChatGPT or Cursor?
Yes. Connect Fireworks AI to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the Fireworks AI actions you allow.
How do users connect Fireworks AI?
Users add their Fireworks AI api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
Which Fireworks AI actions can my agent take?
9 in total: 5 read, 4 write and 0 destructive, such as “create Embeddings”. Your policies decide which of them each agent may call.
Can I make my agent read-only in Fireworks AI?
Yes. Allow read actions and deny writes in the Fireworks AI policy. Your agent can still look things up, and any write it attempts is blocked and logged.
Can my own backend call Fireworks AI too?
Yes. The same Fireworks AI connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.
Plug Fireworks AI into your agent.
Your users connect Fireworks AI once, under your brand. Your agent gets 9 actions behind your policies, with every call on the record.
Free to build · MCP + REST · Audit log on every plan