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 q2522879285-source/minimax-h3-prompting-skill-public --skill minimax-h3-promptinggit clone --depth 1 https://github.com/q2522879285-source/minimax-h3-prompting-skill-publicWrote 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/q2522879285-source/minimax-h3-prompting-skill-public/minimax-h3-prompting)<a href="https://agentmods.dev/skills/q2522879285-source/minimax-h3-prompting-skill-public/minimax-h3-prompting"><img src="https://agentmods.dev/badge/skills/q2522879285-source/minimax-h3-prompting-skill-public/minimax-h3-prompting/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/q2522879285-source/minimax-h3-prompting-skill-public/minimax-h3-prompting"><img src="https://agentmods.dev/badge/skills/q2522879285-source/minimax-h3-prompting-skill-public/minimax-h3-prompting.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 | $0.00161 | $0.04085 |
| Opus 5 | $0.00081 | $0.02042 |
| Sonnet 5 | $0.00032 | $0.00817 |
| Haiku 4.5 | $0.00016 | $0.00409 |
Grade A, and why
minimax-h3-prompting 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 12d 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.
How it starts
The opening of the file, as written. The whole thing — 247 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MiniMax H3 Guided Prompting
Role
Act as an H3-native creative guide, then as the last-mile prompt compiler. Lead the user from intent to a suitable H3 form instead of either interrogating them with a questionnaire or immediately dumping a generic prompt.
Own only H3 fit, guided intake, H3 input-mode routing and serialization, input-role allocation, density budgeting, prompt compression, and H3-specific review targets. Keep story, camera, action, VFX, sound, materials, and continuity with their relevant planning layers when those layers are available. Attribute only the H3 behavior documented here to H3.
Always read h3-interoperability-guide.md before compiling a generation-ready prompt; it is the local model-facing compatibility contract for mode selection, field names, field order, reference labels, language, and timing notation. Read validated-findings.md when current limits, evidence strength, or a claimed H3 capability affects the decision. After choosing a task lane, read only that primary lane in guided-task-recipes.md; also read its cross-cutting audio section when an audio file has a tracked reference role.
1. Guide the Brief to a Decision
Maintain a compact backstage H3 Design Card:
desired result / task lane / evidence class
duration + aspect ratio / H3 input mode
subject or product lock / supplied asset roles
exact text + exact audio obligations
dominant state chain / recurring carrier
semantic node count + editorial microshot density / node richness
requested density surface / active endpoint / decision-critical fidelity risk
Route by brief maturity:
- Exploratory — The user gives only a theme, format, or a request to experiment. Offer two or three materially different H3-native directions, recommend one, state its visible payoff and asset burden, and ask at most one choice that changes the whole design. Delay the long final prompt until the user chooses a direction or asks to write it.
- Partially specified — Fill low-risk defaults, present one compact blueprint containing the dominant state chain, density tier, recurring carrier, and reference plan, then ask only for one unresolved high-impact choice. Carry every approved choice forward.
- Concrete or direct command — When the brief already contains the subject, purpose, duration, key copy/assets, or the user directly asks for the prompt, compile the full prompt and carry every settled decision forward.
- Rendered output supplied — Inspect the video before rewriting; preserve successful layers, identify the earliest causal failure, and use a one-variable repair grounded in the rendered result.
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.
- 12d ago First seen · 247 lines · 161 tokens per session scan A 071fa92b5453
minimax-h3-prompting is a skill published in the GitHub repository q2522879285-source/minimax-h3-prompting-skill-public (11 stars, last pushed 1mo ago), licensed MIT. It adds 161 tokens to every session and 4,085 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
minimax-worldbuilder
Compose MiniMax-H3 video-with-native-audio prompts in H3's mandatory structured format — full-reference (Ref2VA) and text-to-video (T2VA) only. Use whenever the target model is MiniMax-H3: the user says MiniMax, MiniMax-H3, H3, ref2va / r2v / t2v, names a minimaxh3 checkpoint, or names the MiniMaxH3ReferenceToVideo /…
ai-video-prompt-generator
Generate AI video prompts from clean structured prompt specs, validate prompt specs, run an isolated OpenAI-based prompt generator, and lint prompt outputs for viewpoint, action, reference, dialogue, negative prompt, audio prompt, safety tags, and legacy-contamination failures. Use when producing or repairing…
universal-video-prompt-skill
Write one model-agnostic video prompt spec, then compile it to whichever video model you can actually call. Use for cross-model prompt work, model comparison matrices, reusing one brief across providers, or when the target model is not yet available and the work must proceed on another one.
seedance-2-5-prompt-director
A skill for writing and improving prompts for Seedance 2.5, a video-generation system. It covers text-to-video, image-to-video, editing, video extension, references, camera movement, green-screen work, sound, and lip sync.
china-video-prompt-architect
Turn rough China AI video ideas into structured English or Chinese prompt packs, reference-aware motion instructions, shot plans, and focused debugging loops. Use for text-to-video, image-to-video, reference-to-video, video-to-video, product clips, cinematic scenes, transitions, or multi-shot sequences; do not use for…
opensora-video-prompt-architect
Turn rough video ideas into structured OpenSora prompt packs, reference-aware motion instructions, shot plans, and focused debugging loops. Use for text-to-video, image-to-video, video-to-video, product clips, cinematic scenes, transitions, or multi-shot sequences; do not use for unrelated models, account support, or…