$ ~/registry/skill/openai-curated-transcribe

SKILL

transcribe

Name: transcribe
Author: OpenAI

Transcribe audio or video to text with optional speaker labeling.

Stars

★ 25,222

Source

GitHub

Updated

2026-08-29

// securityCaution

Prompt only, no code
Open source
Community-trusted· 22.0k

Summary

Overall risk is low. The skill comes from a highly trusted open-source repository with strong community adoption, but its documentation indicates use of an OpenAI transcription CLI/API, local audio file access, and output generation, so those dimensions are better rated as caution rather than high risk.

CredentialsCaution

Metadata says no key/environment variable is required, but the README explicitly requires OPENAI_API_KEY for live API calls. This is a sensitive credential that could be exposed through local environment handling or logs if misconfigured. The documentation does at least instruct users not to paste the full key in chat, which reduces direct leakage risk.

Network egressCaution

The materials state that transcription is performed 'using OpenAI', indicating that audio content and optional speaker reference audio would be sent to OpenAI services during live use. No unrelated third-party endpoints are declared, and the source repository is trusted, so this is a normal egress pattern consistent with the stated functionality.

Code executionCaution

The README instructs users to run a bundled Python CLI (transcribe_diarize.py) and install the openai package if missing. This means local script execution and possible dependency installation. Such local process execution is a normal tool capability, and the materials do not show red flags such as requests for excessive system privileges.

Data accessCaution

The skill needs to read user-provided audio/video file paths, optional known-speaker reference files, and write results to output/transcribe/ or a user-specified output path. This access scope matches its transcription/diarization purpose, and there is no indication of broad unrelated directory scanning or overprivileged access.

Supply chainLow risk

The source is the open-source GitHub repository openai/skills, with auditable code and strong community adoption (about 22k stars), which are clear positive trust signals. The provided materials do not clearly state the license or maintenance status, but that is not enough to outweigh the low-risk profile of an official, highly trusted source.

Recommendations

Only configure OPENAI_API_KEY when live transcription is actually needed, and inject it securely via local environment variables rather than shared scripts or logs.
Before processing sensitive recordings, confirm that your organization permits sending audio and speaker reference samples to OpenAI services.
Prefer pinned dependency versions and review transcribe_diarize.py plus related docs to reduce installation-time supply-chain uncertainty.
Restrict output paths to a controlled workspace to avoid overwriting or mixing data with other projects.

Audited by: gpt-5.4 · 2026-06-16

// install

Copy the install command and let the AI configure it · recommended for beginners

Please install the "transcribe" skill from askskill:
1. Download https://raw.githubusercontent.com/openai/skills/main/skills/.curated/transcribe/SKILL.md
2. Save it as ~/.claude/skills/transcribe/SKILL.md
3. Reload skills and tell me it's ready

// download

Download SKILL.mdInstall manifest ↗

// docs

Audio Transcribe

Transcribe audio using OpenAI, with optional speaker diarization when requested. Prefer the bundled CLI for deterministic, repeatable runs.

Workflow

Collect inputs: audio file path(s), desired response format (text/json/diarized_json), optional language hint, and any known speaker references.
Verify OPENAI_API_KEY is set. If missing, ask the user to set it locally (do not ask them to paste the key).
Run the bundled transcribe_diarize.py CLI with sensible defaults (fast text transcription).
Validate the output: transcription quality, speaker labels, and segment boundaries; iterate with a single targeted change if needed.
Save outputs under output/transcribe/ when working in this repo.

Decision rules

Default to gpt-4o-mini-transcribe with --response-format text for fast transcription.
If the user wants speaker labels or diarization, use --model gpt-4o-transcribe-diarize --response-format diarized_json.
If audio is longer than ~30 seconds, keep --chunking-strategy auto.
Prompting is not supported for gpt-4o-transcribe-diarize.

Output conventions

Use output/transcribe/<job-id>/ for evaluation runs.
Use --out-dir for multiple files to avoid overwriting.

Dependencies (install if missing)

Prefer uv for dependency management.

uv pip install openai

If uv is unavailable:

python3 -m pip install openai

Environment

OPENAI_API_KEY must be set for live API calls.
If the key is missing, instruct the user to create one in the OpenAI platform UI and export it in their shell.
Never ask the user to paste the full key in chat.

Skill path (set once)

export CODEX_HOME="${CODEX_HOME:-$HOME/.codex}"
export TRANSCRIBE_CLI="$CODEX_HOME/skills/transcribe/scripts/transcribe_diarize.py"

User-scoped skills install under $CODEX_HOME/skills (default: ~/.codex/skills).

CLI quick start

Single file (fast text default):

python3 "$TRANSCRIBE_CLI" \
  path/to/audio.wav \
  --out transcript.txt

Diarization with known speakers (up to 4):

python3 "$TRANSCRIBE_CLI" \
  meeting.m4a \
  --model gpt-4o-transcribe-diarize \
  --known-speaker "Alice=refs/alice.wav" \
  --known-speaker "Bob=refs/bob.wav" \
  --response-format diarized_json \
  --out-dir output/transcribe/meeting

Plain text output (explicit):

python3 "$TRANSCRIBE_CLI" \
  interview.mp3 \
  --response-format text \
  --out interview.txt

Reference map

references/api.md: supported formats, limits, response formats, and known-speaker notes.

Transcribe audio and video into searchable text with speaker diarization support.

—Get→

MCP

★3

Spoken

Search podcasts and fetch clean transcripts with real speaker names.

—Get→

MCP

co.pepys/pepys-mcp

Transcribe audio, video, and podcasts into speaker-labeled timed subtitles.

—Get→

MCP

mcp-transkriptor

Transcribe meetings, media, and URLs into searchable text, summaries, and exports.

—Get→

$ loading_

Please install the "transcribe" skill from askskill: 1. Download https://raw.githubusercontent.com/openai/skills/main/skills/.curated/transcribe/SKILL.md 2. Save it as ~/.claude/skills/transcribe/SKILL.md 3. Reload skills and tell me it's ready

// docs

Audio Transcribe

Transcribe audio using OpenAI, with optional speaker diarization when requested. Prefer the bundled CLI for deterministic, repeatable runs.

Workflow

Collect inputs: audio file path(s), desired response format (text/json/diarized_json), optional language hint, and any known speaker references.
Verify OPENAI_API_KEY is set. If missing, ask the user to set it locally (do not ask them to paste the key).
Run the bundled transcribe_diarize.py CLI with sensible defaults (fast text transcription).
Validate the output: transcription quality, speaker labels, and segment boundaries; iterate with a single targeted change if needed.
Save outputs under output/transcribe/ when working in this repo.

Decision rules

Default to gpt-4o-mini-transcribe with --response-format text for fast transcription.
If the user wants speaker labels or diarization, use --model gpt-4o-transcribe-diarize --response-format diarized_json.
If audio is longer than ~30 seconds, keep --chunking-strategy auto.
Prompting is not supported for gpt-4o-transcribe-diarize.

Output conventions

Use output/transcribe/<job-id>/ for evaluation runs.
Use --out-dir for multiple files to avoid overwriting.

Dependencies (install if missing)

Prefer uv for dependency management.

uv pip install openai

If uv is unavailable:

python3 -m pip install openai

Environment

OPENAI_API_KEY must be set for live API calls.
If the key is missing, instruct the user to create one in the OpenAI platform UI and export it in their shell.
Never ask the user to paste the full key in chat.

Skill path (set once)

export CODEX_HOME="${CODEX_HOME:-$HOME/.codex}"
export TRANSCRIBE_CLI="$CODEX_HOME/skills/transcribe/scripts/transcribe_diarize.py"

User-scoped skills install under $CODEX_HOME/skills (default: ~/.codex/skills).

CLI quick start

Single file (fast text default):

python3 "$TRANSCRIBE_CLI" \
  path/to/audio.wav \
  --out transcript.txt

Diarization with known speakers (up to 4):

python3 "$TRANSCRIBE_CLI" \
  meeting.m4a \
  --model gpt-4o-transcribe-diarize \
  --known-speaker "Alice=refs/alice.wav" \
  --known-speaker "Bob=refs/bob.wav" \
  --response-format diarized_json \
  --out-dir output/transcribe/meeting

Plain text output (explicit):

python3 "$TRANSCRIBE_CLI" \
  interview.mp3 \
  --response-format text \
  --out interview.txt

Reference map

references/api.md: supported formats, limits, response formats, and known-speaker notes.