Connect ChatGPT and Codex to LogRocket to investigate session replays, network requests, issues, and product analytics. The plugin uses the Agent Plugins 1.0 format, connects to LogRocket's hosted MCP server, and includes a skill that guides the agent through LogRocket investigations.
plugins/logrocket/: the self-contained plugin (plugin.json,mcp.json, skill, logo, and usage documentation). See its README for connection options..agents/plugins/marketplace.json: a repository marketplace pointing to the plugin.scripts/package.py: a dependency-free ZIP builder for the plugin.docs/testing.md: manual test cases for a connected plugin.
From the repository root, register it as a marketplace:
codex plugin marketplace add .Open the app's Plugins directory, select LogRocket Plugins, and install LogRocket. Reload the app if the marketplace is not visible. Complete LogRocket OAuth through the connection flow, then test in a new chat. Host support for local marketplaces can vary; see the OpenAI plugin guide.
Run with Python 3.9 or later from the repository root:
python3 scripts/package.pyThis creates dist/logrocket-<version>.zip containing only the plugin files and license, with manifests at the archive root. The version comes from plugins/logrocket/plugin.json. Pass an output path as the first argument to write the archive elsewhere.
Edit files in plugins/logrocket/, bump version in plugins/logrocket/plugin.json for releases, rebuild, and reinstall or refresh the plugin through the host. The skill describes tool behavior but does not freeze tool input schemas; the agent discovers current schemas from the server at runtime.
References:
The code and documentation in this repository are available under the MIT License. The LogRocket name and logo are trademarks of LogRocket, Inc. and are not licensed under the MIT License.