
Ppt Agent
Run a gated, subagent-driven workflow from brief to HTML slides for pitch decks, trainings, and product presentations without the main agent drafting deliverables itself.
Overview
ppt-agent is an agent skill most often used in Build (also Validate, Launch) that runs a gated P0–P5 workflow with subagents to produce HTML presentation decks from briefs, docs, or themes.
Install
npx skills add https://github.com/sunbigfly/ppt-agent-skills --skill ppt-agentWhat is this skill?
- Fixed canonical main chain P0 → P1 → (P2A|P2B) → P3 → P3.5 → P4 → P5 with Gate exit=0 before advancing
- Main agent orchestrates harness, subagents, and validators—formal outputs (outline, style.json, slide-N.html) must come
- Triggers on explicit PPT/slides/deck language and implicit intent (roadshow, boss report, beautify existing slides)
- Branch lock: P2A and P2B are mutually exclusive after entry
- WAIT_USER / WAIT_AGENT hard stops and optional manual audit breakpoints configured in Step 0 interview
- Canonical main chain P0 through P5 with mandatory Gate exit=0
- Mutually exclusive research branches P2A or P2B
- Formal artifacts must be subagent-generated only (search.txt through slide-N.html)
Adoption & trust: 571 installs on skills.sh; 777 GitHub stars; 2/3 security scanners passed (skills.sh audits).
What problem does it solve?
You need a credible multi-page deck but one-shot chat outputs skip research, structure, design consistency, and review gates.
Who is it for?
Solo founders preparing pitch or product decks, PMs turning specs into slide narratives, and anyone asking to beautify or rebuild existing PPT-style content as structured HTML.
Skip if: Single-slide memes, real-time meeting notes only, or teams that refuse subagent latency and want the main agent to draft all slide copy inline.
When should I use this skill?
User asks for PPT, slides, pitch deck, training materials, roadshow deck, beautify/redesign slides, or implicit need for structured multi-page presentation HTML—including 帮我把这篇文档做成 PPT and make a presentation about….
What do I get? / Deliverables
You get validator-gated artifacts through slide HTML and a completed P5 handoff suitable for stakeholder review or further styling in the browser.
- requirements-interview.txt
- outline.txt and style.json
- planning JSON and per-slide HTML files
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Canonical shelf is Build/docs because the primary artifact is structured, multi-page presentation content (HTML slides, outlines, style plans) produced as a ship-ready narrative asset. Deck creation, training modules, and report-to-slide conversion are documentation-style deliverables solo builders need while building and explaining the product.
Where it fits
Turn a startup idea into a gated outline and slide plan before you commit to a landing-page narrative.
Convert an internal spec into styled HTML slides for engineering and design alignment.
Produce a structured planning deck after requirements interview with optional manual audit checkpoints.
Finalize a product intro or keynote HTML deck for demos, webinars, or sales conversations.
How it compares
A multi-step agent workflow with gates and outsourced subagents—not a single-shot slide theme generator or a static PowerPoint template pack.
Common Questions / FAQ
Who is ppt-agent for?
Solo and indie builders, founders, and operators who use Claude Code, Cursor, or similar agents and need structured decks for bosses, customers, trainings, or roadshows.
When should I use ppt-agent?
Use it in Validate when scoping a pitch or landing narrative, in Build when creating training or product explainers, and at Launch when you need presentation-ready materials for distribution—any time you say slides, deck, keynote, or want a document turned into a presentation.
Is ppt-agent safe to install?
Review the Security Audits panel on this Prism page and your repo permissions before enabling harness scripts and subagent network search; treat generated HTML like any user-facing content you will host.
SKILL.md
READMESKILL.md - Ppt Agent
# PPT Agent v4.1 — 主控制台合同 ## 1. 主 Agent 角色 **只做**:维护计划、调用 harness、管理 subagent 生命周期、校验 Gate、与用户交互。 **不做**:代写任何正式产物;手写 subagent prompt;内联执行任何内容生产;用口头判断替代 validator。 **内容生产全量外包红线**:P2A/P2B/P3/P3.5/P4 的所有正式产物(search.txt、source-brief.txt、outline.txt、style.json、planningN.json、slide-N.html 等)**必须且只能**由对应的 subagent 生成。主 agent 自己写出这些产物内容 = 合同违规。主 agent 唯一允许的"写"行为是通过 harness 生成 prompt 文件和通过 validator 校验产物。 ## 2. 全局规则 ### 2.1 步骤控制 - **CLI 固定步骤锁(强制)**:必须严格按 Canonical Plan 的主链 `P0 → P1 → (P2A|P2B) → P3 → P3.5 → P4 → P5` 执行;禁止增删改名。 - 分支二选一:进入 P2A 后绝对不可再跑 P2B,反之亦然。 - **守门规则(Gate)**:进入下个 Step 前,前序 Gate 必须通过;当前步命令执行完毕且 Gate `exit=0` 后才能标记为 `completed`。 - 失败时只允许两种动作:`RETRY_CURRENT_STEP` 或 回退 `ROLLBACK→StepID`。**严禁"跳到后续步骤试试看"**。 - `WAIT_USER` / `WAIT_AGENT` 是硬等待点;未收到输入/FINALIZE 前,**禁止执行后续步骤**。 - **人工审计断点**:是否开启、介入哪些节点、可看哪些材料,必须在 Step 0 采访时写入 `requirements-interview.txt`。断点只能挂在既有主链 Step 内,且只允许主 agent 控制;subagent 不得自行向用户发问。只要 `manual_audit_mode != off`,`review` 完成后的“是否通过人工图审”就是**强制放行点**,主 agent 必须停下来问用户,拿到明确“通过”后才能进入整页终检。 ### 2.2 Subagent 强制调度(核心约束) **通用生命周期**:`create(--model SUBAGENT_MODEL) → RUN(prompt路径) → STATUS… → FINALIZE → close`;完成即关,不复用。Step 4 默认每页先创建一个 PageAgent-N 跑完首轮 Planning → HTML → Review;若用户开启人工审计且在 `review` 放行点未通过,或运行中要求返工,则由主 agent 创建阶段型 PageAgent 或 `PagePatchAgent-N` 继续返工。创建时**必须**显式传 `--model SUBAGENT_MODEL`,禁止省略。`SUBAGENT_MODEL` 由用户在 Step 0 采访时指定(详见 3.1.0 及 6.2)。 **上下文隔离(强制)**:无论 CLI 环境默认是否让 subagent 继承主 agent 上下文,本 skill 要求所有 subagent 必须以**隔离模式**运行——subagent 唯一可见的上下文是主 agent 通过 prompt 文件显式传递的内容。如果 CLI 支持隔离参数(如 `--no-context`、沙箱模式等),必须在《Subagent 操作手册》中记录并在调用模板中包含。主 agent 的对话历史、SKILL.md 内容、环境变量等**不应该**泄露给 subagent。 **Subagent 强制调度表(每行 = 一个必须创建的 subagent)**: | Step | Subagent 类型 | 职责 | 产物 | 主 agent 行为边界 | |------|--------------|------|------|------------------| | P2A | ResearchSynth | 联网检索 + 素材整理 | search.txt, search-brief.txt | 仅 harness 生成 prompt → 创建 subagent → 回收校验 | | P2B | SourceSynth | 用户资料降维整合 | source-brief.txt | 同上 | | P3 | Outline | 大纲构建(含内部自审闭环) | outline.txt | 同上,禁止介入 subagent 内部自审 | | P3.5 | Style | 全局风格锁定 | style.json | 同上 | | P4 | PageAgent-N(每页一个) | 页面规划 + HTML + 审查 | planningN.json, slide-N.html, slide-N.png | 同上,orchestrator 渐进式编排三阶段 | **红线**: - 上表中每个 Step 的产物**只允许对应 subagent 生成**,主 agent 内联生产任何产物 = 合同违规 - 即使 subagent 失败,主 agent 也只能重建 subagent 重跑,不能自己"补写"产物 - 图片模式 `generate` 且用户需要文生图时,额外创建 `ImageGen` 子代理;PageAgent 不承担文生图 - 若用户在人工审计断点提出改单,尤其是在 `review` 后强制放行点给出“不通过”,主 agent 也**必须**通过阶段型 PageAgent 或 `PagePatchAgent-N` 返工;默认从 `review` 重开,让 subagent 继续图审 + HTML 修复;严禁主 agent 直接手改正式产物 **自适应调用协议(每个业务节点强制执行)**: 主 agent 到达上表任意 Step 时,必须按以下流程显式组装 subagent 调用命令: 1. **回查** Section 3.1.1 输出的《Subagent 操作手册》,取出其中的**调用模板**(模型槽位使用 `SUBAGENT_MODEL`) 2. **变量替换**:将模板中的 `{{SUBAGENT_NAME}}`、`{{PROMPT_PATH}}`、`{{MODEL}}` 替换为当前步骤的实际值(`{{MODEL}}` = `SUBAGENT_MODEL`) 3. **显式输出**:将组装后的完整命令输出到对话中(不是脑内执行,是显式写出来) 4. **执行**:按输出的命令执行 subagent 创建、RUN、轮询、回收 禁止“依据操作手册创建”这种含糊引用;必须显式展示组装结果。 ### 2.3 Prompt 生成 - 所有 subagent prompt 必须通过 `prompt_harness.py` 从模板生成;禁止手写 - 所有 `{{VAR}}` 必须填充,残留即 ERROR;输出固定落 `OUTPUT_DIR/runtime/` - 模板/playbook 仅通过 `--inject-file` 注入;主 agent 不手动预读正文 - **Step 0 默认强制模板化**:主 agent 必须先通过 `prompt_harness.py` 生成 `OUTPUT_DIR/runtime/prompt-interview.md`,再依据渲染结果向用户发问;采访运行时模板必须按能力在 `tpl-interview-structured-ui.md` 与 `tpl-interview-text-fallback.md` 之间二选一,不得退化成随手写的一小段简陋问题。 - **Step 0 优先结构化采访 UI**:只要当前 CLI 提供任何等价于 `AskUserQuestion` / `request_user_input` 的原生提问能力,主 agent 就必须优先使用;能力判断看是否支