build-zoom-scribe-app — Codex command: what it does, how to install
Sets up a Zoom Scribe transcription workflow for uploaded or stored media files. It configures ingestion, Build-platform authentication, and callback handling.
~/.codex/prompts/build-zoom-scribe-app.mdWhat build-zoom-scribe-app does
The command configures a backend pipeline to transcribe recorded media through Zoom Scribe. It starts with a preflight audit of the codebase, inspecting existing storage, task queues, and webhooks. Based on latency needs, it determines whether fast mode or batch mode fits the project.
During implementation, Build-platform authorization logic stays separated from transcript processing. The prompt guides adding minimal ingest, submission, parsing, and persistence paths. For batch jobs, it explicitly wires up status polling or callback endpoints.
This workflow handles only stored or pre-recorded audio and video. If your architecture demands real-time transcription during live meetings, Scribe is the wrong fit; switch to RTMS or a bot workflow instead.
What it helps with
- Implement Zoom Scribe audio transcription for stored media
- Select between fast mode and batch mode processing
- Separate Zoom Build credentials from transcript business logic
- Wire webhook callbacks or polling endpoints for batch jobs
Example prompts for the agent
- /build-zoom-scribe-app configure an ingestion pipeline for audio files using batch mode
- /build-zoom-scribe-app add a backend worker to submit media to Zoom Scribe and parse results
- /build-zoom-scribe-app inspect current queueing logic and connect Zoom Scribe callbacks
How to install build-zoom-scribe-app
Manually: put the files in place
~/.codex/prompts/build-zoom-scribe-app.mdCall it with /prompts:build-zoom-scribe-app. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.claude/commands/build-zoom-scribe-app.mdCall it with /build-zoom-scribe-app. For all projects use ~/.claude/commands/
Manually: put the files in place
.github/prompts/build-zoom-scribe-app.prompt.mdManually: put the files in place
.gemini/commands/build-zoom-scribe-app.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about build-zoom-scribe-app
How to install build-zoom-scribe-app in Codex?
Save the command prompt file to ~/.codex/prompts/build-zoom-scribe-app.md. Codex will recognize it immediately.
What does build-zoom-scribe-app do?
It guides the setup of a Zoom Scribe transcription service for uploaded files, helping write ingestion, authentication, and callback code.
Does build-zoom-scribe-app support live meeting audio?
No, it only supports uploaded or pre-stored recordings. For live meeting media, use /build-zoom-rtms-app or /build-zoom-bot.
Source text
build-zoom-scribe-app.md · MIT
--- description: Implement a Zoom Scribe transcription pipeline for uploaded or stored media with the right processing mode and callback flow. --- # Build Zoom Scribe App Use this command when the repo needs a Zoom Scribe transcription workflow for uploaded or stored media rather than live in-meeting media. ## Preflight 1. Inspect the codebase for existing ingestion, storage, queueing, webhook, or transcript-processing code. 2. Confirm the media source, expected latency, language requirements, and whether fast mode or batch mode fits the workflow. 3. Identify the runtime that will own Build-platform credentials, request submission, and callback or polling behavior. 4. Check for required credentials, storage paths, and callback endpoints without printing secrets. ## Plan Before making changes: - state the media source and chosen Scribe processing mode - list the files that will be changed - state the auth path, ingestion path, and transcript output destination - state how the first transcription flow will be verified ## Commands 1. Add or correct the transcription request path in the backend or worker layer. 2. Keep Build-platform auth handling separate from transcript business logic. 3. Add the minimum ingest, submit, parse, and persist flow required for a working transcription pipeline. 4. If batch mode is used, wire the callback or job-status path explicitly. 5. Reuse existing queueing, storage, and observability patterns where possible. 6. Avoid presenting live meeting media workflows as Scribe if the use case really belongs on RTMS or a bot pipeline. ## Verification 1. Re-read the auth handling, submission path, and transcript parsing code after changes. 2. Run local build or tests where available. 3. Verify the chosen processing mode and transcript output path are coherent. 4. State any remaining blocker such as credential audience, media format, callback path, or size constraints. ## Summary ```text ## Result - Action: implemented or updated a Zoom Scribe transcription workflow - Status: success | partial | failed - Details: processing mode, files changed, auth path, verification run ``` ## Next Steps - Test the pipeline with one real media file in a controlled environment. - Add retries, batching, or downstream AI processing only after the base path works. - If the workflow needs live media instead of stored files, switch to `/build-zoom-rtms-app` or `/build-zoom-bot`.
© Zoom · MIT · openai/plugins