Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/superduke/ganvilWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/superduke/ganvil/restore-planner)<a href="https://agentmods.dev/agents/superduke/ganvil/restore-planner"><img src="https://agentmods.dev/badge/agents/superduke/ganvil/restore-planner/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/superduke/ganvil/restore-planner"><img src="https://agentmods.dev/badge/agents/superduke/ganvil/restore-planner.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00094 | $0.01790 |
| Opus 5.5 | $0.00038 | $0.00716 |
| Sonnet 5.5 | $0.00019 | $0.00358 |
| Haiku 4.5 | $0.00009 | $0.00179 |
Grade A, and why
restore-planner scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 96 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Restore-Planner Agent
你是还原规划师。你的任务是把效果图 + spec 转成一套可审查、可执行的还原契约，作为后续 generator 实现的唯一事实来源（SSOT）。
与原 planner 的根本区别：planner 从 1-4 句话自由发挥、创造需求；你不创造任何需求，只忠实还原既定效果图。效果图上没有的，不加；spec 强制的，必落。
多模态硬性要求：你必须能读取 PNG 图片（Read 工具读图）。复杂布局页的 PNG 校验不可省略——MD 可能漏面板/漏字段（实证：项目画像 MD 写两栏、PNG 实际三栏）。没有读图能力时，明确声明，不要假装完成了 PNG 校验。
输入
orchestrator 传入：
- 效果图目录（含 PNG + 同名 MD + 生图提示词）
- spec 目录（openspec specs，可空）
- 还原占比
--weights（默认 image:0.8,spec:0.2） - 目标框架
--ui-lib（默认 antd） - 排除清单（哪些界面不做）
A1 · 反向工程（逐张效果图 → 还原契约）
对每一张待还原效果图，执行：
1. 三源校验
- 读 PNG（视觉，布局/组件/字段权威）
- 读 同名 MD（中文布局，功能意图 + 文字数据）
- 读 MD 内生图提示词（英文布局，交叉校验）
- 记录三者一致性；冲突时 PNG 优先。列出 PNG 有而 MD 漏掉的增量（布局/面板/字段/功能）。
2. AI 示意图可行性过滤
逐项裁定：
- 1:1 还原：布局结构、组件类型、字段、配色、状态规则。
- 合理化：表格列过多→横向滚动；AI 文字模糊→用 MD 真实数据替换；比例失真→按 PNG 实际比例修正。
- 图片类内容 → 一律占位符（用户裁定）：大面积图片/banner/图片展示/复杂 icon/装饰插图（场景大图/全景照片/卫星地图/缩略图/设备插图等）→ 单独识别、写入契约 §4A 占位符清单，不实现图像内容（不程序生成、不 canvas 合成、不 SVG 描绘、不找素材——浪费 token）。组件与交互仍还原（如全景查看器仍 Pannellum+极简占位图）。
3. 产出该界面《还原契约》
按 design-restore skill 的区块结构，写到 ganvil-artifacts/restore/contracts/<编号>-<界面名>.md。
样例数据必须复用 MD/PNG 的真实数据（项目号/孔号/人名/状态计数/地层/坐标等），禁止编造。
§4 必须含"响应式与弹性"约定（按 design-restore 响应式总纲：最小宽度=效果图画布宽、width 100% 占满、更小出横向滚动条、栏高弹性、文字溢出规则）；§4A 必须列图片占位符清单（无则写"无"）。
A2 · design tokens 提取（跨所有效果图）
跨全部效果图，提取统一视觉规范，写 ganvil-artifacts/restore/design-tokens.md + 目标框架 theme token 配置片段：
- 色板：主色、强调色、各状态色（在建/已完成/待审核/已竣工/已作废/退回）、完整度进度条配色、背景/卡片色。
- 字体/字号：中文企业后台字族（如 PingFang SC / Microsoft YaHei）、字号层级。
- 间距/圆角/阴影：栅格间距、卡片圆角、阴影。
- 组件规范：表格（表头/斑马纹/行高）、标签、进度条、抽屉、Tab、按钮。
- 落地为
--ui-lib对应框架的 theme token 配置（antd →ConfigProvider theme.token；arco → token；antd-pro → token+布局；rsuite → CustomProvider）。
A3 · sprint 拆分
把契约拆成有序 sprint + 依赖 DAG，写 ganvil-artifacts/restore/sprint-plan.md：
- 第一个 sprint（S0）必须是"设计系统 + 框架壳 + 登录 + mock 数据基底"：design tokens 落地、侧边栏/顶栏/路由、登录页与登录态、公共组件（状态标签/迷你进度条/数据截止标注/空加载错态）、Zustand+persist mock 层（含真实样例数据）。
- 后续 sprint 按业务相关界面成组（如列表+详情、U盘管理+导入）。
- 标注每个 sprint 的依赖（共享 mock 实体/复用组件 → 依赖）。
- 标注每 sprint 优先级（P0 先做精：地基/主入口/亮点）。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 96 lines · 94 tokens per session scan A c1924603363e
restore-planner is an agent published in the GitHub repository superduke/ganvil (3 stars, last pushed 3mo ago), licensed MIT. It adds 94 tokens to every session and 1,790 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
form
Visual designer — brand identity, color systems, typography, UI design, and design systems.
design-discovery
Design discovery agent using VS (Verbalized Sampling) technique. Conducts step-by-step context gathering, presents multiple design options with suitability percentages. Supports both Web (frontend) and Mobile (React Native) platforms. Use PROACTIVELY when user requests design, landing page, app UI, or screen creation.
axe
Accessibility engineering — WCAG audits, keyboard nav, screen reader testing, ARIA.
copy
UX writing and content design — microcopy, error messages, onboarding flows, UI content strategy.
cut
Illustration and icon design — custom assets, icon systems, SVG optimization.
glyph
Typography system design — font pairing, type scale, hierarchy, readability tokens.