# screen-creative-skills is a category map of 31 screenwriting agents > A Chinese-language collection of agent skills for film and television development, grouped into ten categories that run from novel screening through story analysis, character work, evaluation, creation and workflow orchestration. The interesting structure is the mismatch between the workflow diagram, which draws only the first three categories, and the taxonomy, which is ten deep. **VanGong1999/screen-creative-skills** — Skills for Film and Television Creation Agents: 31 agent skills for evaluating and planning film/TV creation, covering automated AV workflows, vertical short-drama and long-drama IP adaptation, plot/genre extraction, script and novel IP evaluation, character bios, five-element story analysis, and outline generation. - Repository: https://github.com/VanGong1999/screen-creative-skills - Stars: 448 · Forks: 83 - Language: Python - License: MIT - Published: 2026-09-15 · Updated: 2026-09-15 · Language: en - Canonical page: https://hysenlabs.com/projects/vangong1999-screen-creative-skills ## The workflow diagram draws three stages and the taxonomy has ten The panorama is drawn as a left-to-right flow with three boxes: novel screening with four skills, then story analysis with eight, then character analysis with two. Arrows connect them in that order. Below it, the classification runs to ten categories: novel screening, story analysis, character analysis, evaluation and selection, creation and planning, series analysis, workflow orchestration, result processing, tool assistance, and knowledge retrieval. Only the first three appear in the flow. So the diagram is a summary of the front half of the process, and the remaining seven categories are where evaluation, generation and output formatting live. Reading the two together is the only way to know what the collection actually does, and the diagram alone undersells the evaluation half, which is where four dedicated skills live. ## The README points at three sibling repositories under a different owner Three links sit near the top of the page, and none of them is this repository. There is a script knowledge base, a film and television creation skills collection, and a film and television creation agents project, all under the GongLingRui account on GitHub rather than the account this repository lives under. That is a fork or a re-publication arrangement rather than a documentation link, and it has a practical consequence: if you are trying to establish whether a skill has been updated, you have three candidate repositories and no indication from this page which one is canonical. The English counterpart is at least visible in the tree. Alongside category/, docs/ and the licence, the repository holds an english-version/ directory, so a translated set exists somewhere in the history, but the top-level README does not say whether it is current or who maintains it. ## Vertical short drama is the only format with its own create-and-evaluate loop Five content formats are tabled with their typical shape: vertical short drama at one to three minutes per episode across eighty to a hundred episodes, mid-length drama at ten to twenty minutes per episode across twenty to forty, and long-form series or television at forty to sixty minutes per episode across thirty to eighty. Film scripts and IP adaptation sit in the same table without episode counts. Vertical short drama gets the deepest treatment, with drama-planner for emotional-value analysis, three-act structure and target audience positioning, drama-creator for full script generation with dense payoff points and controlled dialogue, and drama-evaluator for adaptation potential, payoff density and difficulty of adaptation. That trio is the only complete loop in the collection. Everywhere else the creation category is empty, so a mid-length or long-form project gets analysis and evaluation skills but no generation skill. ## script-evaluator is the one skill that appears in four of the five formats Reading the content-type table by skill rather than by format shows which ones are general and which are specialised. script-evaluator is listed for mid-length drama, long-form series and film scripts. It is described as evaluating a script along three axes, thought, artistry and watchability, marking strengths and weaknesses explicitly and emitting improvement suggestions. By contrast, drama-evaluator is scoped to vertical short drama alone, assessing short-drama fit, payoff density and predicted adaptation difficulty. That asymmetry is the collection's clearest statement about its market: the long-form formats get a general evaluator and the vertical format gets a specialist one. For IP work the pattern repeats with three different evaluators, ip-evaluator for network information and overall value, novel-evaluator for adaptation potential, and story-outline-evaluator for outline completeness, adaptation feasibility and predicted commercial value. ## Two skills truncate text under different names novel-truncator appears in the novel screening category, described as intelligently truncating text while preserving content integrity, with semantic-completeness preservation, smart breakpoint detection and long-text friendliness. text-truncator appears in the tool assistance category with almost the same description: intelligent truncation preserving semantic integrity, semantic boundary detection, sentence-level completeness. And text-splitter sits next to it, splitting text into fixed-size chunks with smart chunking and context preservation. So the collection carries two near-identical truncation tools under different names, in different categories, and a splitter beside them. Nobody is hurt by it, but it is a sign the categories grew by addition rather than by design, which is worth knowing when you are trying to work out whether a missing capability exists or is simply named something unexpected. ## Two orchestrator skills are the only things here that call other skills Most skills in this collection are single-purpose: read a text, produce an analysis. Two are not. drama-workflow is described as the orchestrator for the dramatic-function analysis of plot points, coordinating the whole process with automatic text segmentation, parallel analysis scheduling and intelligent result merging. Its stated use is deep analysis of long text. plot-workflow produces both major plot points and detailed plot points in one pass, doing layered plot organisation, automatic deep expansion and complete report output, and applies to all formats. Both of these are the skills that make the collection more than a folder of prompts. Without them, running the eight story-analysis skills over one novel is eight separate conversations; with them it is one. That is why the workflow diagram stops at three categories, because the orchestration that would span the rest does not exist for evaluation and creation. ## Version 2.1.0 is dated January while the last push is dated May The version line in the README reads 2.1.0 with an update date of 2026-01-11. The repository metadata shows a last push of 2026-05-26, and there are no GitHub releases at all. So the documented version and the commit history have drifted apart by four months, and there is no tag that tells you which state the 2.1.0 text describes. If you are pinning anything, pin the commit, not the version string. The claimed benefits are worth reading with the same scepticism. A ten-fold efficiency improvement from novel selection to script creation is asserted without a measurement, and the promise that even a beginner can produce a professional-grade script is a claim about quality rather than about tooling. What is concrete is the claim that each stage can be invoked on its own, which is a statement about the architecture rather than about the results. ## Conclusion Adopt screen-creative-skills if your screenwriting work is the Chinese-language market and you want the vertical short drama path in particular, since drama-planner, drama-creator and drama-evaluator form the only complete create-then-evaluate loop in the set. Skip it if you write in English, because the collection ships an english-version directory but the documented taxonomy, the content-type table and the skill descriptions are Chinese, and skip it if you expected a tool rather than a catalogue, since there is nothing to install and no orchestration beyond two skills that call other skills. Before you start, read the version line rather than the repository activity, since the README states version 2.1.0 dated 2026-01-11 while the last push is 2026-05-26 and there are no releases to anchor against. Also resolve which repository you actually want, because this one points at three sibling projects owned by someone else. ## FAQ ### How many skills does screen-creative-skills contain? Thirty-one, stated in the README header. The classification below it runs to ten categories, with named counts of four for novel screening, eight for story analysis, two for character analysis, four for evaluation and selection, two for creation and planning, one for series analysis, two for workflow orchestration, two for result processing, two for knowledge retrieval, and a tool assistance group whose last row reads as still expanding. ### Which skills cover vertical short drama in screen-creative-skills? Three: drama-planner for emotional-value analysis, three-act structure and audience positioning, drama-creator for full script generation with controlled dialogue and dense payoff points, and drama-evaluator for short-drama fit, payoff density and adaptation difficulty. Vertical short drama is the only format with a complete create-then-evaluate set. ### Does screen-creative-skills work outside Chinese-language screenwriting? The repository contains an english-version directory at the top level, alongside category/, docs/ and the licence. The documented taxonomy, the content-type table and the skill descriptions are written in Chinese, and the collection is aimed at vertical short drama, mid-length drama, long-form series and film scripts in that market. ### Which skills in screen-creative-skills orchestrate other skills? Two. drama-workflow coordinates plot-point dramatic-function analysis with automatic text segmentation, parallel scheduling and result merging, and plot-workflow produces major and detailed plot points together with a complete report. Every other skill is single-purpose and does not call the rest of the collection. ## Sources - [Issues](https://github.com/VanGong1999/screen-creative-skills/issues) - [License: MIT](https://github.com/VanGong1999/screen-creative-skills/blob/master/LICENSE) - [README](https://github.com/VanGong1999/screen-creative-skills/blob/master/README.md) - [VanGong1999/screen-creative-skills on GitHub](https://github.com/VanGong1999/screen-creative-skills) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/vangong1999-screen-creative-skills