Converra support

Get help connecting your production agent or coding-agent client to Converra. Converra fixes the behavior, not the code: diagnose a failure, test a bounded change against the current agent, approve what goes live and inspect the production verdict. This page and the setup guides are public; you do not need to sign in to read them.

Contact support

Email support@converra.ai for integration, account-access or product-workflow questions.

If you can sign in, the in-app help page is also available. It requires account access; this public support page does not.

For privacy requests, use the contact in our Privacy Policy. For security reports, use security@converra.ai, the address published in the site's footer. Do not include credentials in either message.

Start with the matching guide

• Claude Code, Codex and Cursor setup
• MCP authentication and connection details
• MCP tool reference
• Vercel AI SDK, OpenAI Agents SDK, LangGraph, Mastra and Python integrations
• How optimization works

Include useful evidence, not secrets

Describe the expected behavior, what happened and when it happened, including your timezone. For connection problems, include the client name and version, the public endpoint URL, the HTTP status or redacted error, and the setup step that failed. Say whether you used OAuth or an API key, but never send the key, access token, password or session cookie.

For an existing fix, identify whether you are asking about a candidate test, an approved release or a live production verdict. A passing test or merged pull request is not proof the production failure stopped. Keep unavailable evidence explicit rather than reporting it as zero failures.

Share customer run content only through an approved secure channel after the scope is agreed. Do not email a transcript or attach a database export to make a troubleshooting report more complete. Do not start a new audit, simulation or deployment just to gather support evidence without the required authority.