Metaphor for AI agents
Metaphor is a neural search engine that finds web content by understanding how people describe and share it, rather than by keywords. 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 Metaphor.
Pull content for a search result
Retrieve the full document contents for search results found by ID, a read-only fetch.
Gather sources for a research task
Pull contents from several search matches to compile background research for a summary or report.
Check a source before citing it
Retrieve a document's content to verify a claim before it is referenced in a summary.
1 Metaphor actions, graded by risk.
Every Metaphor action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.
read
1Look things up. Allowed by default.
- metaphor.get_contentsGet document contents by IDs
write
0Create and change things. Allow, or ask the user first.
- No write actions.
destructive
0Delete, cancel or archive. Ask first, or deny outright.
- No destructive actions.
Metaphor in three steps.
- 01Your users connect MetaphorThey add their Metaphor api key on Arc0 Connect, under your brand. It goes straight into the vault.
- 02You set the rulesReads run, and writes like “updates” can wait for the user to approve.
- 03Any agent can actYour agent calls Metaphor through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
await arc0.policies.set('metaphor', { read: 'allow', write: 'ask', destructive: 'deny', }) # Claude Code: the same connection, one URL $ claude mcp add --transport http arc0 \ https://mcp.arc0.ai/u/u_8f2
How Metaphor connects.
Users add their Metaphor 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 Metaphor 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 Metaphor from any agent.
ClaudeChatGPTCursor
Codex
VS CodeOpenAI Agents SDKClaude Agent SDK
Vercel AI SDK
Mastra
LangGraphMore web data and scraping apps.
Metaphor and Arc0, answered.
Can I use Metaphor with Claude, ChatGPT or Cursor?
Yes. Connect Metaphor 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 Metaphor actions you allow.
How do users connect Metaphor?
Users add their Metaphor 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 Metaphor actions can my agent take?
1 in total: 1 read, 0 write and 0 destructive. Your policies decide which of them each agent may call.
Can I make my agent read-only in Metaphor?
Yes. Allow read actions and deny writes in the Metaphor policy. Your agent can still look things up, and any write it attempts is blocked and logged.
Can my own backend call Metaphor too?
Yes. The same Metaphor connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.
Plug Metaphor into your agent.
Your users connect Metaphor once, under your brand. Your agent gets 1 actions behind your policies, with every call on the record.
Free to build · MCP + REST · Audit log on every plan