Hatcut Chat mode — ChatGPT / Codex guide

If you are in the ChatGPT or Codex web app (not the desktop app), stop. Tell the user to open Codex desktop and paste:

Read https://hatcut.com/chat/chatgpt and set up Hatcut for me, then help me cut silences and add Hinglish captions to my talking-head reel.

On the desktop app:

1. Add the hosted MCP:

codex mcp add hatcut --url https://api.hatcut.com/mcp

If the bundled CLI uses a different syntax, register an HTTP MCP server named hatcut at https://api.hatcut.com/mcp.

2. Verify the server lists get_product_info, create_project, read_project, set_segments, analyze_creative, apply_creative, set_heading, suggest_titles, update_descriptions, submit_export, track_export, get_editor_url.

Call get_product_info (no session) for language capabilities, credit-pack summary, and UTM deep links (utm_campaign=es_latam|de|fr|zh|15day_bet|india, utm_content=capcut_alt; es_mx remaps to es_latam).

3. First edit:

- create_project → save sessionToken + editorUrl
- Ask the user to drop the video at editorUrl and transcribe
- read_project → set_segments (kept=1 keep, kept=0 cut)
- analyze_creative with apply=true (title/heading/thumbnail/zoom/transitions)
- Optional: set_heading / set_captions / suggest_titles / update_descriptions to refine
- submit_export + track_export for captioned output
- Return editorUrl and downloadUrl

No tracks/items/motion graphics. No Clerk login. 3 transcribe + 3 captioned-render credits per session.

Index: https://hatcut.com/chat/llms.txt