Building & integrations

Conversation capture

Turn a session into images or a smooth-scrolling video at whatever size is needed.

Paid $15 list, $12 until 2026-12-31 · one skill, yours for good. The outline, the steps and the file open after purchase.

Background

Rebuild any Codex conversation (active, historical, archived — all work) offline into a single screenshot-ready conversation page, then output two kinds of video assets:

• Still frames: sliced by aspect ratio into 1080×1920 (or 16:9 / 4:5 / 1:1) images at 2x pixel density, ready for image posts, cards and cover images.
• Scrolling video: an MP4 that scrolls smoothly through the whole conversation (H.264, 1080×1920 by default), with a title bar, progress bar and eased in/out.

The data is read directly from the local thread_history (it is not a screen capture), so: the window does not need to be in the foreground, occlusion does not matter, and archived conversations export just as well; fonts and layout are identical every time.

Why it is built this way

Two routes exist because they read different things. Render rebuilds the page from thread_history, so the window can sit behind other programs and an archived conversation exports like any other. A shot goes through PrintWindow, which reads the window's own rendering and survives occlusion, while recording runs ffmpeg gdigrab over the window's screen rectangle and therefore captures the desktop composite. A recording needs the window actually occupying that region; a shot does not.

Canvas and export size are separate parameters. --size sets layout and font size, --shot-size and --video-size set pixels, so 4K comes from raising the export size and enlarging the canvas would only enlarge the type. Slices are cut on turn or message boundaries, so a sentence is never split across two images. Acceptance is a gate rather than a note: a run reports success only on non-empty output, sane PNG dimensions and an MP4 of at least 1 second, and manifest.json holds every image's width, height and sha256.

Where it bites

• A recording of the window on top gdigrab records the screen rectangle, so whatever covers that region is what arrives in the file, and a minimized window cannot be recorded at all. PrintWindow is the part that survives occlusion, which is why covered and minimized behave so differently here. Restore the window or fall back to --full before blaming the encoder.
• A search that matches several conversations --search matches titles and first questions, and several hits raise an error that lists the candidates. The way through is list --search for the id prefix, then --thread. --all exists for the case where every match should be exported, and using it as a shortcut is how the wrong conversation ends up in the library.
• A black frame counted as an asset deskcap prints the colour count of what it captured, and a value of 1 or less means a black or empty frame. It appears when the window was minimized or the rectangle was off screen. Switch to --full or take the render route, and never read a file's existence as proof that it holds a picture.
• Exporting a long conversation whole --max-shots 24 and --max-frames 700 are hard caps, and a very long page makes the run capture fewer slices and write a warning. Cutting with --last-turns or --grep is the intended move for anything past 30 seconds. Render time runs about 4 to 5 times the final duration, so plan for it.

When to use it

Capture a Codex conversation (active, historical or already archived) as images or as a smooth-scrolling video, with size and format (png/jpg/webp, mp4/mov/webm/gif) chosen per task, saved into the asset library under the naming rule. Triggers: conversation screenshot, chat screenshot, export chat history as images, scrolling screenshot, record this conversation as a video, long-conversation screen recording, archived conversation, turn historical conversations into assets, export webm/gif, 4K portrait, put a conversation into a video; window screenshot, desktop screenshot, window recording, desktop recording, recording with the mouse cursor, grab a program's screen, add visual verification for computer use.