minimax-h3
Skill4 starsChanged 3 months agoApache-2.0View on GitHub
Adapt Story-Film shot briefs into MiniMax H3 audio-video prompts using the mandatory H3 prompt-writing format plus an appropriate optional style skill for T2VA, I2VA, FL2VA, L2VA, or Ref2VA.
8 other projects use a version of this.
Adopted by
Most recently taken up by
- ryannel/skillstheir own version, first seen Nov 15, 2026
- calesthio/OpenMontagetheir own version, first seen Nov 9, 2026
- OpenMontageInc/OpenMontagetheir own version, first seen Nov 8, 2026
- xxjrq/seedance-minimax-promptstheir own version, first seen Oct 13, 2026
- Gopyr/gopyr-skills-hubtheir own version, first seen Oct 11, 2026
- teskor-hub/minimax-h3-skilltheir own version, first seen Sep 19, 2026
Other versions8 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 8
- calesthio/OpenMontage★ 81ksame name
- OpenMontageInc/OpenMontage★ 81ksame name
- mini-yifan/open-source-video-gen-skill★ 21same name
- teskor-hub/minimax-h3-skill★ 9same name
- xxjrq/seedance-minimax-prompts★ 3same name
- aleiby/claude-skills★ 1same name
- ryannel/skills★ 1same name
- Gopyr/gopyr-skills-hub★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository139 other files it gives its agents.
- 3d-animation-short-generatorSkill
- ace-step-xlSkill
- adapt-screenplaySkill
- asset-approvalSkill
- audience-simSkill
- audio-masterSkill
- batch-recoverySkill
- beat-sheetSkill
- book-planSkill
- brand-promo-video-generatorSkill
- campaign-brandSkill
- campaign-deliverySkill
- character-bibleSkill
- character-simSkill
- comfyui-api-v2Skill
- comfyui-assetsSkill
- comfyui-binding-auditSkill
- comfyui-cliSkill
- comfyui-discoverSkill
- comfyui-handoffSkill
- comfyui-mcpSkill
- comfyui-offline-batchSkill
- comfyui-runSkill
- comfyuiSkill
- comfyui-troubleshootSkill
- comfyui-workflowSkill
- content-repurposeSkill
- context-shardsSkill
- continuity-checkSkill
- co-op-game-intro-generatorSkill
- creative-pressure-testSkill
- creative-production-specSkill
- crew-reviewSkill
- decision-tree-interviewSkill
- delivery-qcSkill
- design-systemSkill
- dialogue-audio-authoritySkill
- dialogue-timing-preflightSkill
- dialogue-voiceSkill
- director-bookSkill
- documented-creative-discoverySkill
- edit-assistSkill
- editorial-packageSkill
- editorial-reconciliationSkill
- editor-project-exportSkill
- edit-planSkill
- evidence-researchSkill
- ffmpegSkill
- film-completenessSkill
- film-masterSkill
- generation-budgetSkill
- generation-model-setupSkill
- generation-packSkill
- generation-workflow-setupSkill
- guided-production-wizardSkill
- h3-prompt-writingSkill
- handdrawn-live-video-generatorSkill
- imagemagickSkill
- kdenlive-exportSkill
- krea-2Skill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
9 rules. 0 are common, 5 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 6 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 13
Every MiniMax H3 prompt uses this stack in order:
- Distinctive: line 45
Use the mode actually implemented by the selected workflow.
- Very specific: line 72
Use stable <Subject N>, <Picture N>, <Video N>, and <Audio N> labels.
- Very specific: line 77
Use stable (S1), (S2), and later speaker IDs and H3 <d>[Language] ...</d> blocks.
- Distinctive: line 78
When the source brief requires visible-dialogue sync, preserve exact speaker, mouth visibility, cut policy, and measured speech duration.
- Distinctive: line 79
Keep reference labels consistent and preserve Story-Film reference authority scopes.
- Very specific: line 84
If a required exact-audio node is absent from live /object_info, block the selected workflow and report the missing capability.
- Distinctive: line 86
Use concrete image and sound descriptions instead of generic quality adjectives.
- Very specific: line 92
Before prompt-qc or generation, run python scripts/minimax_h3_prompt_validator.py --mode <T2VA|I2VA|FL2VA|L2VA|Ref2VA> --duration <seconds> …
“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
MiniMax H3 Prompting
Required skill stack
Every MiniMax H3 prompt uses this stack in order:
- Story-Film canon, continuity, reference authority, approved dialogue/audio, and the source shot brief.
- The selected ComfyUI workflow, which determines the H3 mode and actual runtime capabilities.
skills/h3-prompt-writing/SKILL.md, which is mandatory and controls final H3 field names, keyframe alignment, shot timing, speaker/dialogue syntax, and base-versus-Ref2VA section order.- Zero or one automatically selected H3 style skill from
references/MINIMAX_H3_SKILL_ROUTING.md. Use more than one only when the user explicitly asks for a hybrid. - This Story-Film adapter's project-specific continuity, exact-audio, reference-scope, and capture constraints.
Story-Film runs python scripts/minimax_h3_skill_router.py --text "<brief>" automatically whenever the selected prompt adapter is minimax-h3; the user does not need to call h3-prompt-writing or a style skill separately. An explicit user selection of one of the eight style skills overrides automatic style classification but still returns through h3-prompt-writing, this minimax-h3 adapter, deterministic H3 format validation, and prompt-qc.
A style skill enriches content. It never replaces h3-prompt-writing, never changes the selected workflow, and never overrides Story-Film canon, reference authority, approved-audio authority, or workflow/runtime capability.
Prompt audit metadata
Store these values in the Story-Film prompt file outside the final model prompt:
adapter: minimax-h3
h3_base_skill: h3-prompt-writing
h3_style_skill: <selected-style-skill-or-none>The generation operator sends only the final H3 prompt body to the model.
Select mode
T2VA: text to audio-videoI2VA: start from a supplied first frameFL2VA: connect supplied first and last framesL2VA: converge to a supplied last frameRef2VA: full-reference generation using labeled image, video, and audio references
Use the mode actually implemented by the selected workflow. Do not infer FL2VA/L2VA support merely because a last-frame asset exists.
Base mode output
Read skills/h3-prompt-writing/references/base-en.txt.
Preserve this core order:
integrated_multimodal_descriptionoverall_soundscapenon_diegetic_music
I2VA, FL2VA, and L2VA also require the correct alignment instruction before the core fields. Build a chronological audiovisual timeline. [Shot 1] has no cut timestamp; later cuts use strictly increasing At MM:SS.mmm times.
Ref2VA output
Read skills/h3-prompt-writing/references/ref-en.txt.
Preserve this order:
subject_definitionssummaryretention_analysisdetailed_descriptionoverall_soundscapenon_diegetic_music
Use stable <Subject N>, <Picture N>, <Video N>, and <Audio N> labels. State exactly what is retained from each reference and where it appears.
Story-Film rules
- Preserve dialogue, lyrics, and visible scene text exactly in the requested/source language.
- Use stable
(S1),(S2), and later speaker IDs and H3<d>[Language] ...</d>blocks. - When the source brief requires visible-dialogue sync, preserve exact speaker, mouth visibility, cut policy, and measured speech duration.
- Keep reference labels consistent and preserve Story-Film reference authority scopes. A character atlas is not automatically composition authority and a temporal tail is not identity authority.
- Tie first or last frames to the timeline explicitly in keyframe modes. When
end_frame.requiredis present, end the described timeline in that state and use FL2VA/L2VA only when the selected local workflow actually uses last-frame conditioning. - For Ref2VA continuation, a registered visual-only previous-shot tail may carry temporal continuity. Its audio must be stripped before it becomes a visual reference.
- When approved dialogue audio exists, preserve model-neutral seconds, approved media identity, and the exact approved source. If the selected workflow uses
MiniMaxH3ExactAudioLock, that workflow is generation authority for exact target audio: convert approved start seconds to H3's 24 fps target timeline, feed each independently timed source throughMiniMaxH3TimedAudio, lock the resulting H3 target-audio latent withMiniMaxH3ExactAudioLock, and when the selected workflow includes it, reinforce the same source through coreMiniMaxH3AddGuideat the same target frame. - A prompt/style skill must not replace that path with H3-regenerated dialogue/music, prompt-only timing, post-hoc muxing as a substitute for conditioning, a different workflow, or a reduced speaker/event model.
MiniMaxH3ExactAudioLockuses autogrowing timed-audio inputs; Do not impose an artificial speaker or utterance count. Preserve the workflow's selected overlap, overflow, gain, and exact-audio policies. - If a required exact-audio node is absent from live
/object_info, block the selected workflow and report the missing capability. Do not silently install the node and do not silently choose another workflow. - Preserve source
capture_behavioras visible capture properties rather than inventing camera hardware. - Use concrete image and sound descriptions instead of generic quality adjectives.
- A style overlay must not invent brand facts, dialogue, lyrics, scientific claims, character identity, reference roles, or continuity facts.
- Save under
04_generation/prompts/minimax-h3/<shot-id>.md.
Deterministic format gate
Before prompt-qc or generation, run python scripts/minimax_h3_prompt_validator.py --mode <T2VA|I2VA|FL2VA|L2VA|Ref2VA> --duration <seconds> --prompt-file <final-model-prompt-body>. Pass only the final H3 model-prompt body, not Story-Film audit metadata. A validation error is a blocker. Repair the prompt and rerun the validator; do not submit malformed H3 syntax to ComfyUI.
Done
The prompt file records its H3 base/style routing, all references resolve, the deterministic H3 format gate passes, H3 section order and syntax are correct, and described audiovisual time equals the requested duration.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

