diff --git a/AGENTS.md b/AGENTS.md index 6bc9688..b29abdf 100644 --- a/AGENTS.md +++ b/AGENTS.md @@ -8,7 +8,7 @@ - `fix: handle missing image API key` - `docs: clarify installation steps` - Commit messages, PR titles, changelog entries, and release notes must be written in English. -- Documentation changes that affect README content must update both `README.md` and `README_en.md` in sync. +- Documentation changes must update all existing language versions in sync, including repository README files and the usage documentation site. ## Changelog diff --git a/CHANGELOG.md b/CHANGELOG.md index d5fcc92..d76dc73 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -7,6 +7,8 @@ Release notes are generated from this file. Keep changelog entries in English. ### Documentation - Add a Korean translation of the README (`README_ko.md`) and a language badge linking to it from `README.md` and `README_en.md`. (#90) +- Replace README language badges with a consistent text-based language switcher across all translations. (#91) +- Add complete English and Korean versions of the Docsify usage documentation with localized navigation, search, and pagination. (#91) ## 0.5.5 diff --git a/README.md b/README.md index 51d9b86..c0b501f 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,8 @@ # Codex PPT Skill -[](README_en.md) [](README_ko.md) [](https://ningzimu.github.io/codex-ppt-skill/#/) [](https://clawhub.ai/ningzimu/codex-ppt) [](https://app.clawmama.run/skills/5lak48/hermes?utm_source=github&utm_medium=issue&utm_campaign=skill_outreach_ningzimu_codex_ppt_skill) [](https://github.com/ningzimu/codex-ppt-skill/stargazers) [](https://github.com/ningzimu/codex-ppt-skill/forks) +**简体中文** · [English](README_en.md) · [한국어](README_ko.md) + +[](https://ningzimu.github.io/codex-ppt-skill/#/) [](https://clawhub.ai/ningzimu/codex-ppt) [](https://app.clawmama.run/skills/5lak48/hermes?utm_source=github&utm_medium=issue&utm_campaign=skill_outreach_ningzimu_codex_ppt_skill) [](https://github.com/ningzimu/codex-ppt-skill/stargazers) [](https://github.com/ningzimu/codex-ppt-skill/forks) 一个面向 Codex 的 PPT 生成 skill,也可在 Claude Code、OpenClaw、Hermes Agent 等支持 `SKILL.md` 的 agent 中使用;在这些非 Codex 环境中通常需要配置 `gpt-image-2`、第三方生图 API 或 OpenAI 兼容格式的生图接口。它把文章、报告、论文、课程笔记等内容转换成“整页图片式”的演示文稿:先规划大纲和视觉风格,再生成每页幻灯片图片,最后用本地脚本组装为 `.pptx`。 diff --git a/README_en.md b/README_en.md index cbb9fcd..872866c 100644 --- a/README_en.md +++ b/README_en.md @@ -1,6 +1,8 @@ # Codex PPT Skill -[](README.md) [](README_ko.md) [](https://ningzimu.github.io/codex-ppt-skill/#/) [](https://clawhub.ai/ningzimu/codex-ppt) [](https://app.clawmama.run/skills/5lak48/hermes?utm_source=github&utm_medium=issue&utm_campaign=skill_outreach_ningzimu_codex_ppt_skill) [](https://github.com/ningzimu/codex-ppt-skill/stargazers) [](https://github.com/ningzimu/codex-ppt-skill/forks) +[简体中文](README.md) · **English** · [한국어](README_ko.md) + +[](https://ningzimu.github.io/codex-ppt-skill/#/en/) [](https://clawhub.ai/ningzimu/codex-ppt) [](https://app.clawmama.run/skills/5lak48/hermes?utm_source=github&utm_medium=issue&utm_campaign=skill_outreach_ningzimu_codex_ppt_skill) [](https://github.com/ningzimu/codex-ppt-skill/stargazers) [](https://github.com/ningzimu/codex-ppt-skill/forks) A Codex skill for generating PowerPoint decks. It can also be used in Claude Code, OpenClaw, Hermes Agent, and other agents that support `SKILL.md`; these non-Codex environments usually require configuring `gpt-image-2`, a third-party image API, or an OpenAI-compatible image generation endpoint. It turns articles, reports, papers, course notes, and other source materials into image-based presentations: first plan the outline and visual style, then generate each full-slide image, and finally assemble the images into a `.pptx` file with a local script. diff --git a/README_ko.md b/README_ko.md index 998d502..1f7cd6f 100644 --- a/README_ko.md +++ b/README_ko.md @@ -1,6 +1,8 @@ # Codex PPT Skill -[](README.md) [](README_en.md) [](https://ningzimu.github.io/codex-ppt-skill/#/) [](https://clawhub.ai/ningzimu/codex-ppt) [](https://app.clawmama.run/skills/5lak48/hermes?utm_source=github&utm_medium=issue&utm_campaign=skill_outreach_ningzimu_codex_ppt_skill) [](https://github.com/ningzimu/codex-ppt-skill/stargazers) [](https://github.com/ningzimu/codex-ppt-skill/forks) +[简体中文](README.md) · [English](README_en.md) · **한국어** + +[](https://ningzimu.github.io/codex-ppt-skill/#/ko/) [](https://clawhub.ai/ningzimu/codex-ppt) [](https://app.clawmama.run/skills/5lak48/hermes?utm_source=github&utm_medium=issue&utm_campaign=skill_outreach_ningzimu_codex_ppt_skill) [](https://github.com/ningzimu/codex-ppt-skill/stargazers) [](https://github.com/ningzimu/codex-ppt-skill/forks) PowerPoint 덱을 생성하는 Codex용 skill입니다. Codex 외에도 Claude Code, OpenClaw, Hermes Agent 등 `SKILL.md`를 지원하는 다른 에이전트에서도 사용할 수 있으며, 이런 비(非)Codex 환경에서는 보통 `gpt-image-2`, 서드파티 이미지 API, 또는 OpenAI 호환 이미지 생성 엔드포인트 설정이 필요합니다. 이 skill은 글, 리포트, 논문, 강의 노트 등의 원본 자료를 "한 페이지 통이미지" 형식의 프레젠테이션으로 변환합니다. 먼저 개요와 시각 스타일을 기획하고, 각 슬라이드를 전면 이미지로 생성한 뒤, 마지막에 로컬 스크립트로 이미지들을 `.pptx` 파일로 조립합니다. diff --git a/docs/_navbar.md b/docs/_navbar.md new file mode 100644 index 0000000..4f66424 --- /dev/null +++ b/docs/_navbar.md @@ -0,0 +1,4 @@ +- 语言 + - **简体中文** + - [English](/en/) + - [한국어](/ko/) diff --git a/docs/en/README.md b/docs/en/README.md new file mode 100644 index 0000000..6fb5483 --- /dev/null +++ b/docs/en/README.md @@ -0,0 +1,61 @@ +# Codex PPT Skill Documentation + +Codex PPT is a presentation-generation skill built for Codex. It also works with agents that support `SKILL.md`, including Claude Code, OpenClaw, and Hermes Agent. It turns articles, reports, papers, course notes, or rough ideas into image-based presentations: first planning the outline and visual style, then generating each complete slide as an image, and finally assembling the images into a `.pptx` file. + +## How to Read This Documentation + +If you just want to get started quickly, begin with the [Quick Start](/en/quickstart.md). + +If you need to install the skill, configure models, or connect it to different agents, see [Installation and Configuration](/en/installation.md). + +If you want to understand the complete generation process, confirmation checkpoints, and quality control, see the [Standard Workflow](/en/workflow.md). + +If you are already using the skill and have encountered a problem, see the [FAQ](/en/faq.md). + +## Pages + +- [Quick Start](/en/quickstart.md): The shortest path for first-time users, example commands, and output files. +- [Design Philosophy](/en/design.md): Why the skill uses image-based presentations, staged confirmation, and a two-skill design. +- [Installation and Configuration](/en/installation.md): Installation and update methods for Codex, OpenClaw, Claude Code, and Hermes Agent, plus API/CLI fallback configuration. +- [Standard Workflow](/en/workflow.md): The complete process, from confirming the outline, style, backend, and sample slide to generating and assembling the full deck. +- [Styles and Personal Style Library](/en/styles.md): Previews of the 12 built-in styles, reproducing a style from reference materials, and saving styles you like to your personal style library for long-term reuse. +- [FAQ](/en/faq.md): Common questions about editability, API keys, sample slides, inserting source materials, and revising individual slides. +- [Example Prompts](/en/prompts.md): Reusable prompts for turning articles into presentations, thesis defenses, executive reports, specified styles, and single-slide revisions. + +## Key Features + +- Image-based presentation generation: Each slide is a complete 16:9 slide image, ideal for scenarios that require strong visual expression and a consistent style. +- Staged confirmation workflow: Confirm the outline, visual style, image-generation method, and sample slide before generating the full deck to reduce rework. +- 12 built-in styles: Options include clean professional, scientific defense, hand-drawn technical explanation, McKinsey-style, party and government red, and teaching courseware. See [Styles and Personal Style Library](/en/styles.md). +- Style matching from reference materials: The skill can review a user-provided presentation, PDF, or screenshots, understand the visual style of each page, and generate a similar style. +- Personal style library: Styles you like can be saved to `~/.codex-ppt-skill/references/`, outside the skill installation directory. They remain available after skill updates and can be reused by name in future projects. +- Place specified source materials on designated slides: You can assign original paper figures, experiment results, architecture diagrams, or screenshots to specific slides. +- Multiple agent environments: In addition to Codex, the skill works with agents that support `SKILL.md`, including Claude Code, OpenClaw, and Hermes Agent. +- Automatic PowerPoint assembly: The skill generates `outline.md`, individual slide images, and `speech.md`, then assembles everything into a `.pptx` file. +- Third-party API support for text models and the `gpt-image-2` image-generation model. +- Companion speaker notes: Speaker notes are generated and inserted into the PowerPoint notes pages by default. +- Targeted post-generation edits for specific slides that need improvement. See the [FAQ](/en/faq.md). + +## Use Cases + +- Turning technical articles into presentation decks +- Turning papers, research reports, or survey materials into presentations +- Turning course notes into teaching slides +- Product introductions, business reports, and project summaries +- Thesis defenses, project applications, midterm reviews, and final acceptance presentations +- Image-based presentations that require a consistent visual language + +## Important Notes + +Codex PPT generates image-based presentations. They offer strong visual consistency, but the text, charts, and shapes on a slide cannot be edited individually as they can in a traditional PowerPoint file. + +If you need to convert the result into an editable presentation, you can use [image-to-editable-ppt-skill](https://github.com/ningzimu/image-to-editable-ppt-skill) after generation. + +You cannot use this skill without access to the `gpt-image-2` image-generation model. See [Installation and Configuration](/en/installation.md). + +## Related Links + +- GitHub repository: https://github.com/ningzimu/codex-ppt-skill +- ClawHub page: https://clawhub.ai/ningzimu/codex-ppt +- Use case showcase: https://github.com/ningzimu/codex-ppt-skill/issues/34 +- Editable presentation conversion skill: https://github.com/ningzimu/image-to-editable-ppt-skill diff --git a/docs/en/_navbar.md b/docs/en/_navbar.md new file mode 100644 index 0000000..67cbfc9 --- /dev/null +++ b/docs/en/_navbar.md @@ -0,0 +1,4 @@ +- Languages + - [简体中文](/) + - **English** + - [한국어](/ko/) diff --git a/docs/en/_sidebar.md b/docs/en/_sidebar.md new file mode 100644 index 0000000..6dc4894 --- /dev/null +++ b/docs/en/_sidebar.md @@ -0,0 +1,8 @@ +* [Documentation Home](/en/README.md) +* [Quick Start](/en/quickstart.md) +* [Design Philosophy](/en/design.md) +* [Installation and Configuration](/en/installation.md) +* [Standard Workflow](/en/workflow.md) +* [Styles and Personal Style Library](/en/styles.md) +* [FAQ](/en/faq.md) +* [Example Prompts](/en/prompts.md) diff --git a/docs/en/design.md b/docs/en/design.md new file mode 100644 index 0000000..a48edd6 --- /dev/null +++ b/docs/en/design.md @@ -0,0 +1,41 @@ +# Design Philosophy + +Codex PPT Skill reflects some of my thinking about using AI to create presentations. + +The most important thing about AI-generated presentations is not speed, but having a controllable process that produces usable results. + +That is why I split the process into several steps: + +1. Read the article, paper, report, or Markdown file +2. Generate an outline so the user can confirm the slide count, titles, and focus of each slide +3. Confirm the overall visual style +4. Generate one sample slide to check whether the text, layout, and style are suitable +5. After the sample is approved, generate the complete image-based presentation slide by slide +6. Assemble the slides into a `.pptx` file and add the script to the notes for convenient presentation reference + +## Why Generate an Image-Based Presentation First? + +Not every scenario requires editability. For technical talks, course notes, book-sharing sessions, paper reviews, or less formal reports, a presentation made from full-slide images is often sufficient and offers better visual consistency. + +If presentation generation and editable presentation conversion are tightly coupled from the beginning, the process becomes heavy. While the outline, style, and sample slide are still being revised, adding editable-structure reconstruction makes the workflow slower, more expensive, and harder to control. + +That is why I split the process into two skills: + +- One generates high-quality image-based presentations +- The other converts an image-only presentation into an editable presentation when needed + +I believe it is more reliable to agree on the content and visual direction first, confirm that the generated results are under control, and only then decide whether editable conversion is necessary. + +## Improvements Based on Community Feedback + +This skill also includes several improvements based on community feedback: + +- Supports the image-generation capability included with Codex subscriptions +- Also supports image-generation models through APIs +- Works in agent environments including Codex, Claude Code, and OpenClaw +- Supports parallel generation with sub-agents to speed up multi-slide presentation creation +- Includes guidance for multiple built-in styles, so you can get good results without writing advanced prompts +- Supports using reference images or existing presentations to generate a new presentation in a similar style +- Preserves user-provided source images such as paper figures, experiment result charts, and screenshots wherever possible, reducing the risk of AI redrawing them incorrectly + +My goal is not to produce a 20-slide presentation with one click. It is to make AI presentation creation feel more like a real workflow: establish the structure, choose the style, approve a sample, then generate the complete deck and inspect the results. diff --git a/docs/en/faq.md b/docs/en/faq.md new file mode 100644 index 0000000..3c80344 --- /dev/null +++ b/docs/en/faq.md @@ -0,0 +1,67 @@ +# FAQ + +## Q: Is the Generated Presentation Editable? + +No. Codex PPT generates image-based presentations in which every slide is a complete slide image. The visual result is more consistent, but the text, charts, and shapes on a slide cannot be edited individually as they can in a traditional PowerPoint file. After generation, if you need to adjust the content or visuals, the recommended approach is to continue in the same conversation and ask the AI to revise the specific slide. + +Unless necessary, do not convert the presentation into an editable format unless you have sufficient Codex credits. + +If you need an editable presentation, try [image-to-editable-ppt-skill](https://github.com/ningzimu/image-to-editable-ppt-skill) after generation. However, `image-to-editable-ppt-skill` is currently experimental. It currently supports only ChatGPT Plus / Pro subscribers using Codex and does not support other agents. If its results are unstable, wait for future update announcements. + +Note: The image-to-editable skill currently consumes a large number of tokens. It is not friendly to Plus users, while Pro users can use it more freely. + +## Q: Does It Support Other Agents? + +Yes. Codex PPT is a `SKILL.md`-based skill. In addition to Codex, it works with agents that support `SKILL.md`, including Claude Code, OpenClaw, and Hermes Agent. + +This skill is developed and tested primarily with Codex, so Codex offers the best compatibility and is recommended. + +Different agents have different image-generation capabilities and tool interfaces. Outside Codex, you will usually need to configure `gpt-image-2` or a third-party OpenAI-compatible image-generation API. See [Installation and Configuration](/en/installation.md), and let the AI guide you through setup based on your current environment. + +## Q: How Do I Update the Skill to the Latest Version? + +Rerun the installation command to overwrite the installed skill with the latest version, or ask your agent to update it for you, then restart the agent. API key configuration and your personal style library are stored outside the skill installation directory and are preserved during updates. See [Installation and Configuration](/en/installation.md) for the commands. + +## Q: Why Does the First Sample Slide Look Good While Later Slides Look Worse or Use a Different Style? + +This usually means the sample slide's style was not passed consistently to later slides, or the prompts, backend, or sub-agent execution method changed during generation. + +Under normal conditions, once a sample slide is approved, later slides should inherit the same visual language, including the color palette, typographic character, layout density, illustration approach, and image-generation backend. If later slides clearly drift, ask the AI to check whether every slide received the sample as a style reference, whether the same image-generation method was recorded and reused, whether a sub-agent switched to another generation method without permission, and whether individual slide prompts are too broad. + +Do not regenerate the entire deck. First select one or two slides with obvious drift and ask the AI to regenerate them using the approved sample as a reference. Explicitly require it to "preserve the sample slide's visual style, use the same image-generation backend, and maintain the same layout density." + +## Q: What Should I Do If the Generated Presentation Looks Bad? + +First confirm that your `gpt-image-2` model can generate images correctly. If the model is unavailable, returns abnormal quality, or a third-party API or relay service does not correctly support `gpt-image-2`, the output quality will be significantly worse. + +After confirming that the model works, ask the AI to address specific issues such as inconsistent styling, text that is too small, overcrowded layouts, unattractive colors, or illustrations that do not match the topic. It is best to regenerate one sample slide first and continue with the full deck only after you are satisfied. + +## Q: What Should I Do If the Generated Slide Images Are Blurry? + +Codex subscribers use the built-in image-generation tool by default. Its generated images have a relatively low resolution, which currently cannot be specified manually. For higher-resolution images, use the `gpt-image-2` API instead—the API/CLI fallback—and configure the API key, base URL, and model name. See [Installation and Configuration](/en/installation.md). + +With the API/CLI fallback, the script defaults to 2K resolution in a 16:9 landscape format. If the image is still blurry, especially on slides with substantial text, ask the AI to generate it at 4K resolution. + +## Q: Why Confirm the Outline First? + +Because most presentation rework comes from structural problems. Confirming the slide count, titles, order, and key points first prevents you from discovering that the content direction is wrong only after every image has been generated. + +## Q: Why Generate a Sample Slide First? + +The sample slide establishes the visual baseline for the complete presentation. After it is approved, later slides can inherit the same color palette, typographic character, density, and visual language. + +## Q: Do I Need to Configure an API Key in Codex? + +If Codex's built-in image-generation tool is available, you usually do not need to configure an API key. You need to configure `OPENAI_API_KEY`, an optional `OPENAI_BASE_URL`, and a model name only when you choose the API/CLI fallback. + +## Q: Can I Insert Original Paper Figures or Architecture Diagrams? + +Yes. During the outline stage, specify the slide number and purpose for each image, and state whether it is strict source material or a style reference. For paper figures, experiment result charts, screenshots, and architecture diagrams, preserve the original information, labels, axes, values, and arrow relationships wherever possible. + +## Q: What Should I Do If One Slide Looks Bad? + +Revise only that slide first. Tell the agent the specific issue—for example, the text is too small, the hierarchy is unclear, the color palette is unsuitable, the visual is overcrowded, or a concept is represented inaccurately. + +## Q: Can I Save My Own Styles? + +Yes. Give the agent presentation screenshots, a PDF, or a complete presentation that you like and ask it to analyze the style. Once you are satisfied with the generated result, ask the agent to save the style to your personal style library at `~/.codex-ppt-skill/references/`. The library is stored outside the skill installation directory, so it is preserved when the skill is updated or reinstalled. If a personal style shares a name with a built-in style, the personal style takes precedence. See [Styles and Personal Style Library](/en/styles.md). diff --git a/docs/en/installation.md b/docs/en/installation.md new file mode 100644 index 0000000..42768d5 --- /dev/null +++ b/docs/en/installation.md @@ -0,0 +1,98 @@ +# Installation and Configuration + +## One-Sentence Installation + +The recommended approach is to send the following sentence directly to Codex and let it install the skill for you: + +```text +Please install this codex-ppt skill for me: https://github.com/ningzimu/codex-ppt-skill +``` + +## Manual Installation for Codex + +Run the following command to install the `codex-ppt` skill in Codex's global skills directory: + +```bash +npx -y skills@latest add ningzimu/codex-ppt-skill \ + --skill codex-ppt \ + --agent codex \ + --global +``` + +Restart Codex after installation so the new skill takes effect. + +You can also download `codex-ppt-skill-v*.zip` from [GitHub Releases](https://github.com/ningzimu/codex-ppt-skill/releases), extract it, place the included `codex-ppt` folder at `~/.codex/skills/codex-ppt`, and restart Codex. + +If you are developing this repository locally, you can symlink the skill directory into the Codex skills directory for real-time testing: + +```bash +mkdir -p ~/.codex/skills +ln -s /path/to/codex-ppt-skill/skills/codex-ppt ~/.codex/skills/codex-ppt +``` + +## OpenClaw Installation + +```bash +openclaw skills install codex-ppt +``` + +If you use OpenClaw's skill allowlist, add `codex-ppt` to the allowlist. + +## Claude Code / Hermes Agent + +Claude Code: + +```bash +npx -y skills@latest add ningzimu/codex-ppt-skill \ + --skill codex-ppt \ + --agent claude-code \ + --global +``` + +Hermes Agent: + +```bash +npx -y skills@latest add ningzimu/codex-ppt-skill \ + --skill codex-ppt \ + --agent hermes-agent \ + --global +``` + +Common destination directories are `~/.claude/skills/codex-ppt` for Claude Code and `~/.hermes/skills/codex-ppt` for Hermes Agent. During local development, you can likewise use a symlink instead of copying the directory. + +## Updating the Skill + +The recommended approach is to send the following sentence directly to your agent and let it update the skill for you: + +```text +Please update the codex-ppt skill to the latest version. The repository is: https://github.com/ningzimu/codex-ppt-skill +``` + +For a manual update, rerun the installation command above for the relevant agent. This overwrites the installed skill with the latest version. Alternatively, download the latest `codex-ppt-skill-v*.zip` from [GitHub Releases](https://github.com/ningzimu/codex-ppt-skill/releases), extract it, and replace the existing `codex-ppt` directory. Restart the agent after the update. + +Updates are safe: runtime configuration such as API keys is stored in `~/.codex-ppt-skill/.env`, while your personal style library is stored in `~/.codex-ppt-skill/references/`. Both are outside the skill installation directory, so updating or reinstalling the skill will not remove them. See the [Releases page](https://github.com/ningzimu/codex-ppt-skill/releases) or the repository's `CHANGELOG.md` for the changes in each version. + +## Image-Generation Model Configuration + +You cannot use this skill without access to the `gpt-image-2` model. The skill depends heavily on the `gpt-image-2` image-generation model. + +## How Do I Know Whether I Have Access to `gpt-image-2`? + +- If you subscribe to ChatGPT Plus or Pro, you have access to the `gpt-image-2` model by default. Codex includes a built-in image-generation tool. +- If you access Codex through a third-party relay API, ask it to generate an image containing complex Chinese text, such as a poem written in running script. Check whether the image is generated successfully and whether the Chinese characters contain errors. If everything works, no further configuration is required. +- If neither option works, you will need to purchase access to a relay API that provides the `gpt-image-2` model. + +You usually do not need to configure the image-generation model manually. While you use Codex PPT, the AI automatically detects the image-generation backend. If none is available, it will ask you to configure an image-generation backend API and guide you through the setup. + +## Notes on Third-Party APIs + +This skill includes a script compatible with OpenAI's official image-generation interface. If you use a third-party `gpt-image-2` relay API, try providing: + +- The relay service's base URL + - If the relay service gives an endpoint such as `https://xxx/v1/images/generations`, set the base URL to `https://xxx/v1`. + - If the relay service already gives `https://xxx/v1`, do not append another layer, which would produce `.../v1/v1`. + - For official OpenAI, `OPENAI_BASE_URL` can be omitted; the default is the official `https://api.openai.com/v1`. +- The relay service's API key +- The exact `gpt-image-2` model name used by the relay service + +After providing this information to the AI, ask it to generate an image. If it still does not work, the relay service may use a custom image-generation scheme that is not fully compatible with the OpenAI image API. Send the relay service's official image-generation documentation to the AI so it can learn the interface and adapt the image-generation script. diff --git a/docs/en/prompts.md b/docs/en/prompts.md new file mode 100644 index 0000000..b5df4f8 --- /dev/null +++ b/docs/en/prompts.md @@ -0,0 +1,45 @@ +# Example Prompts + +## Turn an Article into a Presentation + +```text +Please use the codex-ppt skill to turn /path/to/article.md into a Chinese presentation of about 10 slides. The audience is an internal technical team. +``` + +## Turn a Paper into a Thesis Defense Presentation + +```text +Please use the codex-ppt skill to turn this paper into a 12-slide Chinese thesis defense presentation. Use a scientific defense style. The structure should cover the research background, method, experiments, results, limitations, and conclusion. Preserve the model architecture diagram and the main experiment result figures from the paper on the corresponding slides. +``` + +## Specify a Visual Style + +```text +Please use the codex-ppt skill to generate a presentation. Use the PDF I uploaded as the visual style reference: large titles, generous whitespace, a black-white-gray palette, and small red accents, with the overall feel of a business magazine feature. Review every page image in the material in detail to understand its style, then generate the presentation in a similar style. +``` + +## Specify a Built-in Style + +```text +Please use the codex-ppt skill to turn this material into a 10-slide presentation. Use the built-in "Hand-Drawn Technical Explanation" style, with hand-drawn lines, structured diagrams, lightweight annotations, and clear concept breakdowns. It should be suitable for explaining technical principles and popularizing knowledge. +``` + +You can also replace the style name with another built-in option such as "Clean Professional," "Scientific Defense," "Data Dashboard," "E-Ink Magazine," or "Creative Magazine." + +## Specify the Slide Count and Source Materials + +```text +Please use the codex-ppt skill to turn this material into a 15-slide presentation. Use architecture.png on slide 4 and benchmark-results.png on slide 9. +``` + +## Revise One Slide + +```text +Slide 6 is too dense, and its title is not clear enough. Regenerate only slide 6, preserve the style of the complete presentation, and reorganize the content into a clearer three-part structure. +``` + +## Save a Style to the Personal Style Library + +```text +I really like the visual style of this presentation. Save it to the codex-ppt personal style library for future reuse. Include the color palette, typographic character, layout rules, illustration/chart style, and suitable use cases in the description. +``` diff --git a/docs/en/quickstart.md b/docs/en/quickstart.md new file mode 100644 index 0000000..ac9c26b --- /dev/null +++ b/docs/en/quickstart.md @@ -0,0 +1,35 @@ +# Quick Start + +## Who This Is For + +This page is for first-time Codex PPT users. All you need is an article, report, outline, paper, or set of course notes. Then ask your agent to use the `codex-ppt` skill to generate a presentation. + +## Shortest Way to Get Started + +First install the skill by following [Installation and Configuration](/en/installation.md). Then use the skill directly in Codex to create your presentation. + +```text +Please use the codex-ppt skill to turn /path/to/article.md into a Chinese presentation of about 10 slides. +``` + +If you already know the intended style and use case, be more specific: + +```text +Please use the codex-ppt skill to turn this technical article into a 12-slide Chinese presentation. Use a clean, professional style suitable for an internal technical talk. Slide 5 must use the architecture diagram I provided, and slide 8 must preserve the experiment results figure. +``` + +## Recommendations for Your First Use + +- Ask the agent to generate `outline.md` first, then confirm the slide count, titles, and key points for each slide. +- Do not skip sample-slide confirmation. Review one slide before generating the full deck. +- If one slide is unsatisfactory, revise only that slide instead of regenerating the entire deck. +- If you have a reference presentation, screenshot, or PDF, ask the agent to analyze its style before generating the new presentation. + +## Generated Outputs + +You will usually receive: + +- `outline.md`: Presentation outline +- `origin_image/slide_XX.png`: Final image for each slide +- `speech.md`: Speaker notes for each slide +- `{presentation-name}.pptx`: Final PowerPoint file diff --git a/docs/en/styles.md b/docs/en/styles.md new file mode 100644 index 0000000..6a3c733 --- /dev/null +++ b/docs/en/styles.md @@ -0,0 +1,59 @@ +# Styles and Personal Style Library + +Codex PPT draws its visual styles from two sources: the **built-in styles** shipped with the skill and your **personal style library**, which is stored locally and remains intact when the skill is updated. + +## Built-in Styles + +The skill includes 12 built-in style references, so you can get started without knowing how to write prompts. When creating a presentation, simply name the style you want. For example: + +```text +Please use the codex-ppt skill to turn this material into a 10-slide presentation using the built-in "Hand-Drawn Technical Explanation" style. +``` + +| Clean Professional | Creative Magazine | +| --- | --- | +|  |  | +| E-Ink Magazine | Data Dashboard | +|  |  | +| Retro Flat Illustration | Hand-Drawn Technical Explanation | +|  |  | +| Hand-Drawn Whiteboard | Warm Handmade | +|  |  | +| Scientific Defense | McKinsey-Style | +|  |  | +| Party and Government Red | Teaching Courseware | +|  |  | + +A style is a visual system—including color palette, typographic character, layout density, and illustration language—not a fixed template. Within one style, each slide's layout changes according to its content role, so the slides do not all look identical. + +## Reproducing the Style of Reference Materials + +If the built-in styles do not meet your needs, provide a style reference you like: one screenshot, several screenshots, or a complete presentation or PDF. Ask the agent to analyze the colors, layout, typography, and visual elements before generating a new presentation in the same style: + +```text +Please use the codex-ppt skill to generate a presentation. Use the PDF I uploaded as the visual style reference. Review every page image in detail to understand its style, then generate the presentation in a similar style. +``` + +Note: By default, only the style is reproduced, not the content. Unless you explicitly request it, text and data from the reference material will not be copied into the new presentation. + +## Personal Style Library + +If you are satisfied with a generated presentation's style—whether it is a custom style you developed or a style reproduced from reference material—you can ask the agent to save it for direct reuse: + +```text +I really like the visual style of this presentation. Please save it to my personal style library. +``` + +Key points about how saving works: + +- **Storage location**: Your personal style library is located at `~/.codex-ppt-skill/references/` and can be moved using the `CODEX_PPT_HOME` environment variable. It is stored **outside** the skill installation directory, so updating or reinstalling the skill will not overwrite or remove your personal styles. +- **Automatic discovery**: No registration is required after saving. The next time you choose a presentation style, the agent automatically scans your personal style library and lists your styles alongside the built-in styles. +- **Personal styles take precedence**: If a personal style has the same name as a built-in style, your personal style is used. You can use this behavior to customize a built-in style by saving an adjusted version under the same name. +- **Reuse by name**: In the future, simply name the style—for example, "Generate this presentation using the 'Dark Data Technology' style." + +After generation, if a deck uses a custom or adjusted style, the agent will also mention in its final report that you can save it. There is no need to save an unmodified built-in style again. + +## Related Pages + +- [Example Prompts](/en/prompts.md): Complete prompts for choosing a built-in style, matching a reference style, and saving a style. +- [FAQ](/en/faq.md): What to do when the style drifts or a slide is unsatisfactory. diff --git a/docs/en/workflow.md b/docs/en/workflow.md new file mode 100644 index 0000000..0faed0b --- /dev/null +++ b/docs/en/workflow.md @@ -0,0 +1,69 @@ +# Standard Workflow + +## Overview + +The Codex PPT workflow emphasizes staged confirmation. Instead of immediately generating the entire presentation, it first confirms the outline, style, image backend, and sample slide to reduce rework. + +## Stage 1: Review the Source Material + +The agent first determines: + +- The topic and central argument +- The target audience +- The presentation objective +- The required slide count +- Content that must be included or excluded +- Whether any image assets are required + +## Stage 2: Confirm the Outline + +The agent generates `outline.md`, which usually includes: + +- Slide number +- Slide title +- 3-5 key points per slide +- The role of each slide, such as cover, agenda, concept explanation, process, comparison, data evidence, or summary +- Optional visual ideas +- Required image assets and how they will be used + +No final slide images, `speech.md`, or `.pptx` file should be generated before the outline is approved. + +## Stage 3: Confirm the Visual Style + +The agent proposes 2-3 style directions and recommends one. Candidates come from the 12 built-in styles, including clean professional, scientific defense, hand-drawn technical explanation, McKinsey-style, party and government red, and teaching courseware, as well as your personal style library. The agent can also reproduce a style from screenshots, a PDF, or a presentation you provide. See [Styles and Personal Style Library](/en/styles.md) for complete style previews. + +After a style is selected, the entire presentation should maintain a consistent visual language, while individual slide layouts may vary according to the content. + +## Stage 4: Confirm the Image-Generation Backend + +The Codex built-in image-generation tool is preferred. The workflow switches to `scripts/image_gen.py` only when the built-in tool is unavailable, the user explicitly requests an API/CLI fallback, or the current capability cannot satisfy the request. + +Once the image backend is confirmed, the entire presentation should use the same backend without switching midway. + +## Stage 5: Generate and Confirm a Sample Slide + +Generate one sample slide first and check: + +- Whether the text is clear +- Whether the style matches expectations +- Whether the information density is appropriate +- Whether the colors and layout are stable +- Whether the design can scale to the full presentation + +Generate the full deck only after the sample slide is approved. + +## Stage 6: Batch Generation + +After the sample slide is approved, the agent generates `origin_image/slide_XX.png` one slide at a time. In environments that support sub-agents, one sub-agent generates each slide in parallel to speed up multi-slide production. Every slide follows the same style and image-generation backend approved for the sample. + +## Stage 7: Quality Review and Fixes + +Before assembly, the agent reviews every slide for text clarity, consistency with the outline, clipped content, visual consistency, unnecessary page numbers, and overlapping elements. Slides with serious issues are regenerated using stricter prompts, while minor local issues are preferably corrected with targeted image-editing tools. + +## Stage 8: Speaker Notes and Assembly + +The agent generates the speaker notes in `speech.md`, then uses `assemble_ppt.py` to assemble the presentation into a `.pptx` file. The speaker notes are automatically added to the notes section of each slide. + +## Stage 9 (Optional): Save the Style + +If the presentation uses a custom or adjusted style, the agent notes in the final report that you can save it to your personal style library and reuse it by name in the future. See [Styles and Personal Style Library](/en/styles.md). diff --git a/docs/index.html b/docs/index.html index 06adf75..b5e8dbb 100644 --- a/docs/index.html +++ b/docs/index.html @@ -2,7 +2,7 @@
-