seedance-2-5
Skill2 starsChanged 4 months agoMITView on GitHub
Seedance2.5 skill for AI video production. Use whenever a user mentions Seedance2.5, Seedance 2.5, AI video, short videos, marketing videos, product demos, talking-head videos, video prompts, storyboards, scene cards, B-roll, video repair, or continuity QA. Turn supplied scripts, briefs, assets, footage notes, or approved storyboards into grounded Seedance-ready plans, prompts, and checks; do not invent product claims, fabricate people, or claim platform controls without checking the active Seedance interface.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- OpenMontageInc/OpenMontagetheir own version, first seen Nov 8, 2026
- calesthio/OpenMontagetheir own version, first seen Nov 8, 2026
- Gopyr/gopyr-skills-hubtheir own version, first seen Oct 11, 2026
- xiaoliangliang/seedance-2.5-prompt-skilltheir own version, first seen Jul 31, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- calesthio/OpenMontage★ 78ksame name
- OpenMontageInc/OpenMontage★ 78ksame name
- xiaoliangliang/seedance-2.5-prompt-skill★ 2same 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.
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 21
Read references/video-routes.md for the chosen route.
- Distinctive: line 28
Treat the active Seedance UI/API as authoritative for duration, aspect ratio, audio, editing, and reference-asset controls.
- Distinctive: line 40
Do not force every request through a screenplay workflow.
- Very specific: line 48
Use the structure in references/prompt-patterns.md: references, start state, camera/action, dialogue/sound, end state, and only necessary …
“Common” is measured: kept by at least 6 in 10 of the 3 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
Seedance2.5
Plan and prompt video production from material the user can verify. This skill produces an editable production plan and prompt pack; the active Seedance interface determines the generation controls and renders the final media.
Capabilities
| Route | Minimum input | Deliverable |
| Narrative / short drama | Script or storyboard | Scene ledger, clip plan, continuity-safe prompts |
| AI short video | Topic, hook, or short script | Beat sheet, visual beats, video prompts |
| Marketing / product video | Product brief plus approved claims/assets | Claim-safe visual plan, shot list, prompts |
| Talking-head video | Speech script, transcript, or footage notes | B-roll and cutaway plan, prompts, edit notes |
| Product demo / explainer | Feature facts and product references | Demo sequence, proof beats, prompts |
| Repair / continuation | Failed clip plus the desired correction | Focused repair prompt and preserved state |
Read references/video-routes.md for the chosen route. Read references/route-templates.md to produce a route-specific brief and prompt preamble. Read references/production-contract.md before any multi-clip production task. Read references/prompt-patterns.md while writing or repairing prompts.
Honest boundary
- Preserve supplied facts, dialogue, people, products, and reference assets. Ask rather than fill gaps with plausible fiction.
- Build prompts, plans, and QA; do not claim to render the final video, clone a person, create a voice, or use a platform feature unless the active interface supports it.
- Require authorization for reference images, videos, voices, performers, and brands. Do not imitate protected characters or living performers without permission.
- Treat the active Seedance UI/API as authoritative for duration, aspect ratio, audio, editing, and reference-asset controls.
Workflow
1. Lock the production brief
Record only what changes the output: target audience and platform, source material, approved facts/assets, delivery format, sound policy, and whether the work is draft or locked.
For marketing work, use only claims supported by the user’s brief. For talking-head work, keep the supplied language intact and plan supporting visuals rather than rewriting the speaker’s voice.
2. Select the smallest route
Do not force every request through a screenplay workflow. A 15-second product demo needs proof beats; a talking-head edit needs B-roll timing; a short drama needs a scene ledger. Use the relevant route template and omit unrelated deliverables.
3. Plan before prompting
Make the visible story clear before adding style words. Each clip needs a purpose, an opening state, one playable change, and an ending state. Carry confirmed state forward; do not reset people, props, location, lighting, or product details between clips.
4. Write prompts in production order
Use the structure in references/prompt-patterns.md: references, start state, camera/action, dialogue/sound, end state, and only necessary constraints. Prefer observable direction over vague labels such as “premium” or “cinematic.”
5. QA the pack
Validate the route brief first:
python3 scripts/validate_video_plan.py --route short-video path/to/plan.mdFor a multi-clip pack, run:
python3 scripts/validate_prompt_pack.py path/to/prompt-pack.mdReplace short-video with marketing, talking-head, product-demo, or narrative as appropriate. Resolve errors. Treat warnings as a prompt to manually check whether the next clip has enough inherited state.
Delivery
Return only the route-relevant artifacts: brief, fact/asset list, beat or clip plan, prompt pack, and unresolved assumptions. For multi-clip work, include the validator result.
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.

