Customgpt for AI agents

CustomGPT.ai builds AI chatbots trained on a business's own content, letting a company deploy an assistant that answers from its own documents. 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.

AI and modelsAPI keyMCP + RESTcustomgpt.ai
AUDIT LOG · CUSTOMGPTPOLICY: acme-support
09:41:07 · claude · u_8f2read
customgpt.list_pages
List Agent Documents✓ allowed · 212ms
09:41:08 · claude · u_8f2read
customgpt.get_stats
Get Agent Statistics✓ allowed · 164ms
09:41:09 · claude · u_8f2write
customgpt.create_project
Create CustomGPT Project✓ approved · approved by user
09:41:12 · claude · u_8f2destructive
customgpt.delete_page
Delete Page from Agent✕ blocked · policy: deny
EVERY CUSTOMGPT CALL, ON THE RECORD
01 · USE CASES

What agents do in Customgpt.

01

Send a message to a project's agent

Send a message to a project's chatbot and get back an answer grounded in its sources.

02

Check a message's trust score

Check a response's trust score before relying on it for a customer-facing answer.

03

Confirm before deleting a source

Confirm before deleting a source document, since the agent's answers relying on it would change.

02 · ACTIONS

40 Customgpt actions, graded by risk.

Every Customgpt action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.

read

23

Look things up. Allowed by default.

  • customgpt.get_stats
    Get Agent Statistics
  • customgpt.list_pages
    List Agent Documents
  • customgpt.get_message
    Get Message
  • customgpt.get_plugins
    Get Agent Plugins
  • customgpt.get_project
    Get CustomGPT Project
  • customgpt.list_sources
    List Agent Sources
  • customgpt.list_personas
    List Persona Versions
  • customgpt.list_projects
    List CustomGPT Projects
  • customgpt.get_usage_limits
    Get Usage Limits
  • customgpt.get_user_profile
    Get Current User Profile
  • customgpt.get_page_metadata
    Get Page Metadata
  • customgpt.get_report_traffic
    Get Traffic Analytics Report
  • customgpt.get_project_license
    Get Project License
  • customgpt.get_report_analysis
    Get Analytics Chart Data
  • customgpt.search_team_members
    Search Team Members
  • customgpt.get_project_settings
    Get Project Settings
+ 7 MORE

write

13

Create and change things. Allow, or ask the user first.

  • customgpt.add_source
    Add Source to Project
  • customgpt.update_source
    Update Source Settings
  • customgpt.create_project
    Create CustomGPT Project
  • customgpt.update_project
    Update Project
  • customgpt.create_conversation
    Create Conversation
  • customgpt.update_user_profile
    Update User Profile
  • customgpt.update_page_metadata
    Update Page Metadata
  • customgpt.update_project_license
    Update Project License
  • customgpt.update_project_settings
    Update Project Settings
  • customgpt.reindex_page
    Reindex Page
  • customgpt.clone_project
    Clone CustomGPT Project
  • customgpt.submit_message_feedback
    Submit Message Feedback
  • customgpt.activate_persona_version
    Activate Persona Version

destructive

4

Delete, cancel or archive. Ask first, or deny outright.

  • customgpt.delete_page
    Delete Page from Agent
  • customgpt.delete_source
    Delete CustomGPT Source
  • customgpt.delete_project
    Delete CustomGPT Project
  • customgpt.delete_project_license
    Delete CustomGPT Project License
03 · HOW IT WORKS

Customgpt in three steps.

  1. 01Your users connect CustomgptThey add their Customgpt api key on Arc0 Connect, under your brand. It goes straight into the vault.
  2. 02You set the rulesReads run, writes like “create CustomGPT Project” can wait for the user, and “delete Page from Agent” can be denied outright.
  3. 03Any agent can actYour agent calls Customgpt through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
POLICY.TS
await arc0.policies.set('customgpt', {
  read: 'allow',
  write: 'ask',        // create_project
  destructive: 'deny',  // delete_page
})

# Claude Code: the same connection, one URL
$ claude mcp add --transport http arc0 \
    https://mcp.arc0.ai/u/u_8f2
04 · AUTH AND DATA

How Customgpt connects.

Users add their Customgpt 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 Customgpt 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
05 · WORKS WITH

Use Customgpt from any agent.

Claude
ChatGPT
Cursor
Codex
VS Code
OpenAI Agents SDK
Claude Agent SDK
Vercel AI SDK
Mastra
LangGraph
07 · FAQ

Customgpt and Arc0, answered.

Q01

Can I use Customgpt with Claude, ChatGPT or Cursor?

Yes. Connect Customgpt 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 Customgpt actions you allow.

Q02

How do users connect Customgpt?

Users add their Customgpt 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.

Q03

Which Customgpt actions can my agent take?

40 in total: 23 read, 13 write and 4 destructive, such as “create CustomGPT Project”. Your policies decide which of them each agent may call.

Q04

Can I stop my agent from deleting things in Customgpt?

Yes. Actions like “delete Page from Agent” are graded destructive. Set destructive actions to deny, or to ask so the user approves each one, and blocked calls still show up on the audit log.

Q05

Can my own backend call Customgpt too?

Yes. The same Customgpt connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.

Get started

Plug Customgpt into your agent.

Your users connect Customgpt once, under your brand. Your agent gets 40 actions behind your policies, with every call on the record.

Free to build · MCP + REST · Audit log on every plan