Search skills/

media-use skill

A HyperFrames skill that finds or generates music, sound effects, images, voiceover, avatars, transcripts and colour grades for a video, using HeyGen or free local tools.

in heygen-com/hyperframes Video & media
Our verdict Recommended with caveats

Capable and well tested, but the best paths need a HeyGen login or credits and some local models are heavy installs, so check the setup page first.

5/5 Depth expertise shipped Note Security Review before installing 0d Maintenance Updated today 1.2M Adoption installs OK License Apache-2.0, commercial OK

What the media-use skill does

media-use is the asset layer for HyperFrames projects. One command, npx hyperframes media-use resolve --type <type> --intent "...", finds or makes a media file, saves it under the project's assets folder and records it in a ledger, so renders never fetch anything over the network. The skill documents two routes per need. With HeyGen (CLI signed in via OAuth or an API key) you get a catalog of music and images, HeyGen voices and voice cloning, avatar video and background removal. The free route uses local or open tools: MusicGen for music, 21 bundled offline sound effects plus Openverse, local FLUX for images, Kokoro for text-to-speech, Parakeet or whisper for transcripts, and the CLI's own cutout and colour-treatment commands.

The skill also covers mixing placed audio (fades, ducking, loudness), grading and LUTs for dull or flat footage, caption authoring from transcripts, and a memory file for preferences. It is the largest skill in the repo: about 96 files, roughly 44 scripts and 25 test files, with reference pages for each task. It tells the agent to confirm before a paid call it chose to make, to credit CC BY sounds, and offers fixes for gaps such as silent cuts only once.

Good for

  • Adding music, voiceover, captions and sound design to an agent-built video
  • Getting transcripts, cutouts and colour grades with local tools and no account
  • Teams that already pay for HeyGen and want avatars or catalog music in the loop

Not for

  • Anyone who wants to avoid sending prompts or audio to third-party generation services
  • Machines that cannot run the optional local models (they need RAM and Python setup)

Depth 5/5

A 5 KB SKILL.md backed by about 96 files, including around 44 scripts, 25 test files, reference pages per task, bundled sound effects and LUTs.

Dependencies and costs

  • HyperFrames CLI (Node.js 22+) and FFmpeg
  • heygen CLI with OAuth login or an API key for catalog music, images, HeyGen voices and avatars (credits may apply)
  • Python 3.10+ and kokoro-onnx for local text-to-speech; optional local MusicGen, FLUX or LTX models
  • Telemetry: The skill relies on the HyperFrames CLI, whose usage counters are on by default (linked to your HeyGen account if signed in); disable with npx hyperframes telemetry disable. Resolve calls to HeyGen, Openverse or other providers go to those services; --local-only skips every network provider.

Security review

Review before installing

Skill that finds or makes music, sound effects, images, voiceover, transcripts and grades for HyperFrames videos using HeyGen or free local tools. Its scripts send each provider's own API key to that provider, install Python packages on demand, and send usage events to PostHog unless you opt out; once signed in to HeyGen, those events are linked to your email and username. Only part of the 69 code files was read in full, and the HyperFrames and heygen CLIs were not reviewed.

Worth knowing

  • Sends usage data by default (can be turned off)resolve and the edit tools post usage events to PostHog (us.i.posthog.com) by default; after HeyGen sign-in they are linked to your email and username read from ~/.heygen/credentials. Opt out with HYPERFRAMES_NO_TELEMETRY=1, DO_NOT_TRACK=1 or hyperframes telemetry disable; a one-time notice is printed.skills/media-use/scripts/lib/telemetry.mjs:7
  • Needs an API key for its own serviceHEYGEN_API_KEY / OAuth token goes to the HeyGen API (project .env files up to five folders up are loaded, but a project .env cannot redirect the HeyGen base URL); Gemini keys go to generativelanguage.googleapis.com and ELEVENLABS / OpenAI / Groq keys to their own APIs.skills/media-use/audio/scripts/lib/heygen.mjs:55
  • Makes network requestsDownloads media from HeyGen, Openverse, Google Gemini/Lyria and other providers; media-fetch.mjs refuses non-public hosts and caps redirects and body size.skills/media-use/scripts/lib/media-fetch.mjs:41
  • Installs packages or toolsRuns python -m pip install on demand for google-genai and python-dotenv (Lyria) or transformers/torch/soundfile/numpy (local MusicGen), and the docs add kokoro-onnx and optional local models.skills/media-use/audio/scripts/lib/bgm.mjs:40
  • Runs remote code at an unpinned versionStandalone installs run npx hyperframes media-use ... without a version; plugin installs run npx --yes hyperframes@<plugin version>. pip installs have no version pins.skills/media-use/scripts/resolve.mjs:14
  • Relies on a CLI or package we could not reviewRelies on the HyperFrames npm CLI, the heygen CLI, ffmpeg and optional local model tools whose code is not in the skill folder.skills/media-use/SKILL.md:10
  • Changes files outside your projectWrites an anonymous ID and notice flag to ~/.hyperframes/config.json (mode 0600) and keeps a global media ledger under ~/.media.skills/media-use/scripts/lib/telemetry.mjs:49

What we checked

The skill folder (116 files) on Oct 29, 2026. Not a full review: 1 of 116 files were not downloaded (large repository; we reviewed the skill folders); we read 10 of 69 code files; it relies on code outside the repository: npm package hyperframes (HyperFrames CLI, run through npx); heygen CLI and the HeyGen, Openverse, Gemini, ElevenLabs, OpenAI and Groq services; PyPI packages installed on demand (google-genai, transformers, torch, kokoro-onnx and others) and local model weights.

NVIDIA SkillSpector static scan: 151 pattern hits (MCP Rug Pull 107, Privilege Escalation 12, Data Exfiltration 12, Prompt Injection 10, Tool Misuse 5, Dangerous Code Execution 2). Most hits are documentation or test fixtures: .env mentions are the key-loading logic, DELETE /voices lines are in a mocked test, curl lines are example API calls to ElevenLabs, OpenAI and Groq, and 'npx hyperframes' is the CLI invocation. The PostHog telemetry and on-demand pip installs are real behaviors.

A review, not a certification. Skills change after we check them — read the files before installing.

License

Apache-2.0 — commercial use allowed. Bundled sound effects carry their own credits in CREDITS.md.

Can it run unattended?

Not really. It is a helper for individual asset requests and asks the person to confirm paid calls; it has no autonomous multi-hour loop.

Alternatives

video-shotcraft

Vincentwei1021/video-shotcraft

Remotion-based skill for cinematic product promo videos, with 124 shot recipe cards, a full template, beat sync, sound design and a motion workbench.

With caveats Check first Depth 5/5

Higgsfield AI Skills

higgsfield-ai/skills

Nine skills for generating images, video, brand kits, product shots and websites through the paid Higgsfield service and its CLI.

With caveats Broad access Depth 3/5

Pixel2Motion

nolangz/pixel2motion

A skill that traces a raster logo into a clean SVG, then builds a choreographed logo animation as standalone HTML, with GIF/video export and motion checks.

Recommended Check first Depth 4/5

Vox Director

Alisa0808/vox-director

A skill that turns one topic, a talking-head video or a photo into a finished paper-collage explainer video, using the Atlas Cloud API and local ffmpeg.

With caveats Check first Depth 4/5

FAQ

What is media-use?

media-use is a skill in HeyGen's HyperFrames repo that gets or generates the media a video needs: music, sound effects, images, voiceover, avatars, transcripts, cutouts and colour grades. It saves everything locally with a record so renders stay offline.

How do I install media-use?

It is part of the HyperFrames core set. Run npx hyperframes skills update, or npx skills add heygen-com/hyperframes and select it. For HeyGen features install the heygen CLI and run heygen auth login --oauth; check setup with npx hyperframes media-use resolve --doctor.

Is media-use free?

There is a free path using local models and open sources such as Openverse, Kokoro and whisper, which need no account. HeyGen's catalog, voices and avatars use an OAuth allowance or API credits, and the skill asks before a paid call.

media-use vs a generic text-to-speech or image skill: what is different?

media-use is built around the HyperFrames project: it writes assets into the project's folder, records provenance, and handles mixing and grading for video. A generic skill would not know that structure.

Sources: GitHub repository; install counts from skills.sh via Best Skills (CC BY 4.0). Review based on the published SKILL.md.