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.
npx skills add seed-forge/harness-ai-kit --skill base-manual-capturegit clone --depth 1 https://github.com/seed-forge/harness-ai-kitWrote 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/skills/seed-forge/harness-ai-kit/base-manual-capture)<a href="https://agentmods.dev/skills/seed-forge/harness-ai-kit/base-manual-capture"><img src="https://agentmods.dev/badge/skills/seed-forge/harness-ai-kit/base-manual-capture/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/skills/seed-forge/harness-ai-kit/base-manual-capture"><img src="https://agentmods.dev/badge/skills/seed-forge/harness-ai-kit/base-manual-capture.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00101 | $0.01997 |
| Opus 5 | $0.00051 | $0.00999 |
| Sonnet 5 | $0.00020 | $0.00399 |
| Haiku 4.5 | $0.00010 | $0.00200 |
Grade A, and why
base-manual-capture 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 9d 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 — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Base Manual Capture
把"打开系统、逐步操作、逐步截图、逐步确认"沉淀为域中立、工具中立的采集基座。本 Skill 只负责采集——产出截图与结构化操作记录；手册正文由 pg-manual-builder 一类写作层技能编写。
何时使用
- 需要为一个可访问系统（有地址+测试账号）生成带截图的操作手册
- 需要按功能描述逐步走查系统并与用户确认业务流程/概念是否符合预期（B 阶段）
- 需要刷新旧手册的截图（系统版本变化后重采）
- 需要把已确认的操作路径固化为可回放脚本（C 化）
本 Skill 不负责：写手册正文、判断业务规则是否正确（由用户在 B 阶段确认）、桌面端采集（当前仅 Web 通道）。
前置输入
- 系统访问地址
- 测试账号（口令仅用于登录，不落台账/记录，见脱敏规则）
- 功能描述（口述或简要均可）——要采哪条/哪几条流程
- 手册目标读者（影响截图取舍与注意事项详略）
- 输出目录：
project_root/manual_root（与pg-manual-builder约定一致；截图落manual_root/captures/<flow>/）
若功能描述模糊，先请用户给一句话流程目标（如"用户登录后创建工单并提交审核"），再开采。
工具解析优先级
按以下顺序选用任一可用的浏览器自动化工具，任一可用即可执行：
- Playwright MCP（首选，可精确 selector + 稳定截图）
- browser-use MCP（会话式操作，Qoder 等自带）
- 本地 playwright CLI（
.agents/tools/ms-playwright或playwright命令）
开采前先做一次探活：能打开目标地址并成功登录即视为通道可用；都不可用则停止并提示用户配置。
采集通道
capture_channel: web（当前唯一实现）- Client/桌面端为预留扩展位，本版不实现；未来新增通道时复用同一套台账/记录契约
B 阶段主流程（核心）
SOP-B0 采集范围确认（人工确认点，必做）
开采前先锁定“手册要覆盖哪些功能”，避免漏采：
- 登录系统后先做一次功能盘点：遍历导航菜单/功能入口，列出发现的功能模块与子页面清单（含数量，如“综合查询含 5 类 20+ 子表单”）
- 把清单交给用户勾选确认：哪些进手册（逐页详采）、哪些只截代表页、哪些不覆盖；未经确认不得自行裁定覆盖范围
- 确认结果落为 flow 清单（写入 capture-trace 或独立
flows.md）：每个 flow 一行，标注 待采/已采/跳过 - B 阶段退出条件同步收紧：flow 清单全部处理完毕（而非单条流程走完）才算采集完成，交给写作层前再向用户报一次覆盖对账（已采 N/应采 M）
SOP-B1 环境确认
- 用选定工具打开地址并登录，确认进入目标系统首页
- 固定视口
1920×1080（保证截图尺寸一致，详见截图规范） - 显式确认
project_root/manual_root，建立manual_root/captures/<flow>/目录 - 初始化空的
capture-ledger.md与capture-trace.yaml（字段见契约）
SOP-B2 逐步采集循环（每步都做全）
对流程的每一步：
- 执行一个操作（点击/输入/选择/提交等）——一步只做一个动作
take_screenshot落盘到captures/<flow>/<step_id>-<页面名>.png- 记录四要素：入口（菜单路径/按钮位置）、动作、预期结果、注意事项
- 向用户播报本步理解并请求确认/纠偏：用一两句话复述"我在做什么、看到什么"，等待用户确认或纠正后再进入下一步
- 把该步写入
capture-ledger（含用户确认状态）与capture-trace（含 selector/action 供 C 化）
SOP-B3 分支与异常路径
正常路径走完后，按用户指点补采异常分支（权限不足、校验失败、数据为空、提交失败等），同样逐步截图+确认。异常步在台账中标注所属分支。
SOP-B4 阶段产物落盘
一条流程采完，确保产出齐全：
- 截图文件（
captures/<flow>/） capture-ledger.md（截图台账，四要素+确认状态已填）capture-trace.yaml（操作路径记录，供 C 化）- 草稿手册骨架（可选）：按 flow 生成"章节+步骤占位"的 md，方便用户即时审查
What ships with it
7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 9d ago First seen · 127 lines · 101 tokens per session scan A d54160b40097
base-manual-capture is a skill published in the GitHub repository seed-forge/harness-ai-kit (22 stars, last pushed 8d ago), licensed Apache-2.0. It adds 101 tokens to every session and 1,997 once invoked, about $0.0005 per session on Opus 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 2026-08-30.
Other skills, from other repositories
baoyu-url-to-markdown
Fetch any URL and convert to markdown using baoyu-fetch CLI (Chrome CDP with site-specific adapters). Built-in adapters for X/Twitter, YouTube transcripts, Hacker News threads, and generic pages via Defuddle. Handles login/CAPTCHA via interaction wait modes. Use when user wants to save a webpage as markdown.
ai-web-scraping-scrapegraph
AI-powered web scraping - extract data using natural language prompts.
browser-automation-notte
Browser automation - control browser sessions, scrape pages, and run AI agents.
extract-webpage-data
Extract structured data from web pages using AI.
agent-browser
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
ultimate-browsing
Reaches web pages a plain fetch cannot: renders JS, drives clicks and forms, captures screenshots, holds a login, and gets past WAF blocks through platform-native readers and stealth Chrome. Use for any page work beyond retrieving static text.