Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add sutchan/Agent-Skills-Hub --skill lumagit clone --depth 1 https://github.com/sutchan/Agent-Skills-HubWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sutchan/agent-skills-hub/luma)<a href="https://agentmods.dev/skills/sutchan/agent-skills-hub/luma"><img src="https://agentmods.dev/badge/skills/sutchan/agent-skills-hub/luma/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sutchan/agent-skills-hub/luma"><img src="https://agentmods.dev/badge/skills/sutchan/agent-skills-hub/luma.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00093 | $0.02046 |
| Opus 5 | $0.00046 | $0.01023 |
| Sonnet 5 | $0.00019 | $0.00409 |
| Haiku 4.5 | $0.00009 | $0.00205 |
Grade A, and why
luma scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
86% identical to luma — 19 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
luma
The cinematic video-generation tool skill — define the shot, reference with stills + frames, economize with
drafts, assemble the sequence, and mind the rights. The agent briefs (and can drive the API where connected);
the human judges every clip; WoopSocial publishes. (Ships with tools/integrations/luma.md.)
The POV: shots, not films — and drafts, not gambles
Dream Machine's 2026 edge is per-shot cinematic quality: Ray3 was the first "reasoning" video model (it interprets, generates, self-evaluates, retries), with 16-bit HDR, keyframes, character reference, and Modify (restyle real footage, keep the performance). The top-1% operator holds four truths the marketing softens. (1) Generations are shots, not films: 5–10 seconds each, Extend degrades past the initial clip — a brand film is a shot list, generated shot-by-shot and assembled in an editor with sound added in post (no native audio; that's Veo/Kling territory). (2) Draft first, always: iterate cheap in Draft Mode, Hi-Fi master only the winner — because HDR at 1080p runs ~16× standard credits and monthly credits expire at reset. (3) Variance is real: the same prompt differs run-to-run more than half the time (independent testing) — so client-grade consistency comes from references (still-first image-to-video, character reference on base Ray3, keyframes), never from luck. (4) The rights fine print: free = watermarked non-commercial, the commercial floor is the Plus-level plan (tier tables genuinely conflict — verify in-app), and Luma keeps a marketing license on your generations — the clause client work needs to know about.
Read these first
- ai-video — the model-agnostic router/craft above the video tools.
- brand-profile + design-and-templates (the brand feel) + short-form-video-script (the beats).
The framework: DREAM
(Depth: references/the-dream-framework.md.)
- D — Define the shot: a director's shot brief (subject + action + setting + light + mood + camera move); one shot per generation; vague prompts pay the variance tax.
- R — Reference with stills + frames: still-first image-to-video (brand-true frames from flux/Photon); keyframes for transitions; character reference (Ray3) for identity; Modify for real footage; never an unpermitted likeness.
- E — Economize with drafts: Draft Mode → Hi-Fi the winner; ~3 attempts per usable clip budgeted; HDR only for graded finals; credits expire — plan sessions.
- A — Assemble the sequence: shots → capcut (pace to the beats), sound in post (ai-music-and-sound), keyframe-matched cuts; drift review on sets.
- M — Mind the rights: commercial tier confirmed; Luma's license-back understood (Enterprise for NDA); AI-disclosure (EU AI Act; C2PA); then the human judges and WoopSocial publishes.
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 102 lines · 93 tokens per session scan A abbf5031b574
luma is a skill published in the GitHub repository sutchan/Agent-Skills-Hub (2 stars, last pushed yesterday), licensed MIT. It adds 93 tokens to every session and 2,046 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. It is 86% identical to luma, differing in 19 lines, and is treated as a copy.
Other skills, from other repositories
writing-skills
Use when creating new skills, editing existing skills, or verifying skills work before deployment.
receiving-code-review
Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation.
writing-plans
Use when you have a spec or requirements for a multi-step task, before touching code.
skill-authoring
Author SKILL.md skills: frontmatter, validator limits, structure.
skill-creator
Create, improve, evaluate, benchmark skills. Use when authoring a new skill, updating an existing one, running evals, or optimizing a skill's description for triggering. Don't use for invoking skills, writing prose, or scaffolding Python projects.
iflytek-hyper-tts
A text-to-speech tool that turns written text into an MP3 recording. It can use an authorized voice and adjust speaking speed, volume, and pitch.