Composio docs show OpenAI Codex integrating with Specific MCP for live call analytics

Composio published a developer documentation page on June 2 showing how OpenAI Codex can connect to Specific through the Model Context Protocol, or MCP, for sales-call analysis workflows. The page says Codex users can use natural-language prompts from the terminal to summarize objections from recent calls, identify calls that need follow-up and generate analytics reports for a team. (composio.dev) OpenAI’s Codex documentation says Codex supports MCP servers in both its CLI and IDE extension, including streamable HTTP servers and OAuth authentication. The Composio page is part of the company’s broader toolkit library for connecting AI agents to third-party services. (composio.dev) A separate Composio toolkit page describes Specific as a sales call analytics platform that helps teams understand call performance and says the integration can be used to analyze calls, generate sales insights, track performance trends and automate reporting through natural language. ### What exactly did Composio publish on June 2? Composio’s “How to integrate Specific MCP with Codex” page presents the integration as a way to connect Codex to a Specific account through Composio’s managed MCP server. The page says users can “summarize top objections from last week’s calls,” “list sales calls needing follow-up actions,” and “generate call analytics report for my team” from the terminal. (composio.dev) The documentation also says the Specific MCP server gives an AI agent “structured and secure access” to a Specific account. In the rendered page, Composio lists supported tools including Create Company, Create Conversation, Create or Update Company and Create Or Update User. (composio.dev) ### How does this fit with OpenAI Codex’s MCP support? OpenAI’s Codex documentation says MCP connects models to external tools and context, and that Codex supports MCP servers in both the CLI and the IDE extension. The same page says Codex can connect to streamable HTTP servers, supports bearer-token and OAuth authentication, and stores MCP configuration in `~/.codex/config.toml` or a project-scoped `.codex/config.toml`. (composio.dev) OpenAI’s docs also say users can add and manage servers with the `codex mcp` command. In the terminal interface, users can inspect active MCP servers with `/mcp`, according to the Codex documentation. (composio.dev) ### What do the docs say Codex can do with Specific data? Composio’s page frames the integration around call-review and sales-operations tasks rather than code generation. The examples surfaced in search results focus on three actions: objection summaries, follow-up identification and team analytics reporting. Specific’s toolkit page on Composio describes the underlying service as a way to analyze calls, generate sales insights, track performance trends and automate reporting. (developers.openai.com) That description aligns with the Codex examples shown on the integration page. ### How would a developer set it up? Composio’s instructions say users can add the server in Codex by running a setup command that generates an MCP URL and starts browser-based authentication. The page also says users can authenticate manually with `codex mcp login composio` and confirm the server registration with `codex mcp list`. (composio.dev) The same page says Codex App follows the same pattern as the VS Code setup flow. Composio shows a configuration example using the ` endpoint and an `x-consumer-api-key` header in `.codex/config.toml`. (composio.dev) ### Where can developers check the next step? OpenAI’s Codex MCP documentation says the next step after configuration is managing servers through the `codex mcp` command set or editing `config.toml` directly. Composio’s Specific integration page and Specific toolkit page remain the two published references for the workflow and supported use cases as of June 2. (developers.openai.com) (composio.dev)