Media Use logo

Media Use

Community
waker240
media-use

Resolve reviewed local BGM, sound effects, images, icons, and brand assets into frozen project files plus a manifest. Use when a HyperFrames composition needs non-voice media that is already supplied, staged under assets/, present in the trusted local cache, or explicitly selected from a local review catalog after discovery and audition.

Overview

Publisherwaker240
RepositoryFullVideoProductionSkill
Skill namemedia-use
Stars
211
Forks
47
Bundled files
15
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 15 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by waker240 on GitHub. Read the source before you install it.

Installation

Install the Media Use AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/waker240/FullVideoProductionSkill/tree/main/skills/media-use
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/waker240/FullVideoProductionSkill.git /tmp/FullVideoProductionSkill
mkdir -p .claude/skills
cp -r /tmp/FullVideoProductionSkill/skills/media-use .claude/skills/media-use
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Media Use in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Media Use on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Media Use is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

media-use

Turn an approved local asset into a stable project path and ledger record. This skill is an inventory and ingestion layer: it performs no web search, download, media generation, or voice synthesis.

For revisioned audition decisions, generated images/video, derivatives and the distinction between a reviewed asset and one actually used in a release, read review and derivative provenance. Keep those editorial records alongside the resolver manifest; do not pretend its existing flags implement a full editorial state machine.

Voice is outside this resolver. fish-audio-api is the bundled synthesis adapter. Respect supplied narration or a user-selected provider; prepare that audio through its supported route, then import the resulting local narration file and measured timing into the composition.

Public package setup

No media recordings, author library, review database or cache is included. Supply your own reviewed files and rights records. A new project can begin with an empty library; discovery may return no results. Use explicit --source ingestion when you already have a permitted file.

The project wrapper finds .agents/skills/media-use or the directory named by MEDIA_USE_SKILL_DIR. Its optional MEDIA_CATALOG points to a catalog within the project/workspace; --catalog overrides it. The resolver's optional MEDIA_USE_GLOBAL_CACHE_DIR selects a user-owned cache. These paths are configuration, not assets supplied by this release. Standalone resolver/catalog commands read environment variables; the scaffolded wrapper additionally reads the project .env.

Resolve

bash
node <SKILL_DIR>/scripts/resolve.mjs --type <type> --intent "<description-or-id>" --project <dir>

Types: bgm, sfx, image, icon, and the local-only brand record.

For brand, a local frame.md, FRAME.md, design.md, or DESIGN.md can be frozen as the design record. Flat YAML frontmatter is retained when present; a plain Markdown design document is still a valid local record.

Resolution is deterministic and local:

  1. Reuse a valid project-manifest record matching the requested id, path, exact prompt, or entity.
  2. Adopt a matching unregistered file from the project's assets/ tree, preserving supplied review/rights flags. Strict HyperFrames audio delivery still requires those fields before wiring.
  3. Reuse a frozen record from the local global cache. BGM and SFX cache entries must carry local-origin provenance.
  4. On a miss, stop with staging instructions. The agent or user selects and reviews the asset before retrying.

Review-catalog handoff

In a scaffolded HyperFrames project, npm run audio:discover -- --type bgm --query "<semantic job>" (or --type sfx) searches workspace review catalogs without copying or approving anything. Treat its result as a lead: inspect the source/license/technical metadata, audition the file, then ingest that exact reviewed candidate with --source, --reviewed, and the catalog's --expected-sha256. Preserve source id/page, catalog id, license name/URL, a substantive review note, and creator/attribution when present. Catalog membership alone is never approval, and a render must never reference the review-library path directly.

After explicit ingestion, the resolver prints one concise result:

text
resolved sfx_001 → assets/sfx/soft-whoosh.wav (sfx)

Ingest an explicit local file

Use --source after selecting and reviewing a file. The resolver copies it into .media/, probes its metadata, and records its local provenance.

bash
node <SKILL_DIR>/scripts/resolve.mjs \
  --type sfx \
  --intent "soft interface whoosh" \
  --source /absolute/path/to/reviewed-whoosh.wav \
  --reviewed \
  --expected-sha256 <64-hex-catalog-hash> \
  --source-id <source-provider-asset-id> \
  --source-page <original-asset-page> \
  --catalog-id <catalog-asset-id> \
  --license <license-name-or-id> \
  --license-url <license-page> \
  --review-note "auditioned under narration; clean short tail" \
  --project .

--source is an explicit ingestion path, so it bypasses discovery and cache matching. It accepts a regular local file only. When --catalog-id is present, the review, hash, source, rights, and substantive-note fields shown above are mandatory. The resolver verifies the hash both before and after freezing the copy, before it writes the manifest.

Flags

FlagMeaning
--type, -tbgm, sfx, image, icon, or brand
--intent, -iDescription, manifest id, or manifest path
--entity, -eOptional exact entity match
--sourceReviewed local file to freeze and register
--source-pageOriginal asset page retained as audit metadata
--licenseLicense name or identifier retained as audit metadata
--license-urlLicense page retained as audit metadata
--attributionAttribution text retained as audit metadata
--reviewedRecord that the candidate was explicitly reviewed before ingestion
--expected-sha256Require the source bytes to match a 64-hex catalog hash
--creatorCreator retained as audit metadata
--source-idSource provider's asset id retained as audit metadata
--catalog-idLocal review-catalog asset id retained as audit metadata
--review-noteHuman audition/review note retained as audit metadata
--project, -pProject directory; defaults to .
--adoptInventory supported files already under assets/; bulk inventory alone does not approve audio for rendering
--jsonEmit machine-readable output

Adopt an existing project

bash
node <SKILL_DIR>/scripts/resolve.mjs --adopt --project .

Adoption probes supported media with ffprobe, appends records to .media/manifest.jsonl, and regenerates .media/index.md. Bulk adoption is an inventory operation, not editorial approval: BGM/SFX records still need substantive review and rights provenance before the strict HyperFrames sound gate will allow them to render. Resolve a selected staged audio asset individually with the relevant review/license flags to make that provenance explicit. Narration and voice files are intentionally ignored; TTS outputs remain owned by the Fish Audio workflow.

Inventory and cache

  • .media/manifest.jsonl is the project source of truth.
  • .media/index.md is the readable inventory.
  • ~/.media/ is the frozen cross-project cache. MEDIA_USE_GLOBAL_CACHE_DIR may point the resolver at another trusted local cache root.

The resolver does not purge older cache entries. It simply refuses BGM or SFX records whose provenance does not establish local selection or adoption. Legacy adopted records marked with both provider: local and adopted: true remain reusable; a remote provider marker is never accepted as local provenance.

Cache eligibility is not current editorial approval. Before reuse, check the current project's decisions and exclusions by candidate id and source hash. A cached asset rejected in this project remains excluded even if an earlier project approved it.

Miss handling

When resolution misses, use the HyperFrames project's read-only audio:discover command, or select a properly licensed asset outside this skill, then ingest the reviewed file with --source. Preserve source, license, attribution, hash, catalog identity, and review notes in the project record.

Requirement

ffprobe is recommended for duration, codec, and dimensions. Resolution still works when probing fails, but those metadata fields remain empty.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Media Use AI skill do?

Resolve reviewed local BGM, sound effects, images, icons, and brand assets into frozen project files plus a manifest. Use when a HyperFrames composition needs non-voice media that is already supplied, staged under assets/, present in the trusted local cache, or explicitly selected from a local review catalog after discovery and audition.

Why use Media Use on TypingMind?

Because you install it once and use it with any model. Media Use is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Media Use in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/waker240/FullVideoProductionSkill/tree/main/skills/media-use. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Media Use?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Media Use?

As many as you like. As long as a model supports skills, you can use Media Use with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Media Use AI skill free?

Yes. It is published on GitHub by waker240 under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇