Edit videos with Codex (ChatGPT)

Codex (the agent inside ChatGPT desktop) speaks MCP. Register PandaStudio with one CLI command, then ask Codex to edit your video. PandaStudio ships its own MCP server inside the app, so there's no system-wide Node install required.

Install PandaStudio + ChatGPT desktop with Codex

PandaStudio from writepanda.ai. ChatGPT desktop from openai.com/chatgpt/download. Codex tool use needs a paid ChatGPT plan (Plus, Pro, Team, or Enterprise).

Launch PandaStudio at least once. The first launch writes a local credentials file the MCP server auto-discovers — there's no token to manage.

Put pandastudio-mcp on your PATH (one-time, macOS only)

PandaStudio ships its MCP server binary inside the installed app. On macOS, link it onto /usr/local/bin once:

bash

sudo /Applications/PandaStudio.app/Contents/Resources/cli/install-cli-symlink-mac.sh

That drops pandastudio-mcp (and pandastudio) on your PATH so any external tool can find them.

Windows installer (NSIS) already adds the launcher directory to %PATH% automatically — skip this step.

Register PandaStudio with Codex

Codex ships an mcp add subcommand that registers a stdio MCP server. Run this once:

bash

codex mcp add writepanda -- pandastudio-mcp

Codex writes the server entry to ~/.codex/config.toml automatically. No JSON editing, no env vars, no token.

If codex --version reports "command not found", install the Codex CLI first: npm install -g @openai/codex. The ChatGPT desktop app includes its own bundled Codex runtime, but the codex mcp add command lives in the CLI.

Ask Codex to edit your video

Open ChatGPT desktop, start a new chat, and pick Codex as the agent. Then try:

"Open my latest recording, transcribe it, remove filler words and silences, then export an MP4."
"Create a new 9:16 project, import the file at ~/Recordings/launch.mov, add bold captions, and preview it."
"List all my projects from this week and tell me which one is longest."

Codex picks PandaStudio's tools out of its MCP registry and drives the editor. You stay in the chat; PandaStudio stays open so you can scrub or tweak by hand any time.

Troubleshooting

"codex: command not found" — install the Codex CLI: npm install -g @openai/codex. Then re-run the codex mcp add command from step 3.
"pandastudio-mcp: command not found" — the launcher symlink from step 2 isn't in place. Verify the install: which pandastudio-mcp. On macOS make sure /usr/local/bin is on your PATH (it is by default in zsh and bash).
PandaStudio doesn't appear in Codex's tool list — check the registration with codex mcp list. If writepanda is there but reports errors, run codex mcp test writepanda for diagnostic output. Most common cause: PandaStudio hasn't been launched yet, so the local credentials file doesn't exist.
No PandaStudio installed (CI, server)? — You can still register the npm-published server, which spawns under your system Node: codex mcp add writepanda -- npx -y @writepanda/mcp. Functionally identical, just slower to start and requires Node 18+ on PATH.
Prefer editing the config directly? — Codex stores MCP servers in ~/.codex/config.toml as TOML. The block written by codex mcp add looks like [mcp_servers.writepanda] with command = "pandastudio-mcp". You can edit it there if you need to scope by project or pin a version.

Reference

CLI & MCP commands

Browse all 60+ verb-noun commands the agent can call.