All guides

Setup guides · 2 min read

Recon for Codex CLI

Add Recon's MCP server to OpenAI's Codex CLI. One config block, OAuth flow, then customer memory in every terminal session.

Codex CLI supports MCP servers via its TOML config. Adding Recon is one config block.

1. Open the Codex config

Codex stores config at ~/.codex/config.toml. Create it if it doesn't exist.

2. Add the Recon MCP server

Append this block:

[mcp_servers.recon]
url = "https://askrecon.com/api/mcp"
transport = "http"

3. Authenticate

Run Codex once to trigger OAuth:

codex

Codex will detect the new server and prompt for OAuth. Sign in with your Recon account.

4. Try it

codex exec "Use recon to brief me on Acme before my call."

Or in an interactive session:

> use recon to find at-risk accounts on Enterprise

When this shines

Codex CLI is best for scripted runs and CI-style automation:

  • Pre-merge check: codex exec "Use recon.impact to check if any code in this PR affects an at-risk customer."
  • Nightly report: cron a Codex command that asks Recon for the weekly customer brief and writes it to Slack.
  • Release prep: codex exec "Use recon to summarize what changed this week for our top 20 accounts."

How it works

Codex CLI speaks the same Streamable HTTP MCP transport as Claude Code and Cursor. All six Recon tools are read-only.

Start with a snapshot

Connect your first source in under 10 minutes.

Sign up free. No card required. Your AI tools get customer memory the moment discovery finishes.

Start free