[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-1347":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":13,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":10,"pushedAt":10,"updatedAt":25,"readmeContent":26,"aiSummary":27,"trendingCount":15,"starSnapshotCount":15,"syncStatus":17,"lastSyncTime":28,"discoverSource":29},1347,"site-memory","LittleYier\u002Fsite-memory","LittleYier","Persistent memory for browser automation AI Agents. Learns site knowledge automatically, updates after every visit, and gets better over time.","",null,"JavaScript",250,9,8,0,1,2,3,46.9,"MIT License",false,"main",true,[],"2026-06-12 04:00:09","# site-memory\n\n**English** | [中文](README.zh-CN.md) | [日本語](README.ja.md)\n\nBrowser automation agents forget everything between runs. Same site, same page structure to figure out, every single time. Wasted time, wasted tokens.\n\nsite-memory lets your agent remember sites it has visited and go straight to the goal. \n\nOn the WebVoyager benchmark, it cuts cost by 70% to 90% and makes tasks 4x faster with no loss in accuracy.\n\n- Learns site knowledge automatically\n- Updates after every visit\n- Gets better over time\n\n## Why you need site-memory\n\nWithout site-memory, booking a hotel on Booking.com takes 35 steps and nearly 8 minutes. \nWith site-memory, the second run takes 5 steps and 30 seconds. Stable sites get even faster after a few visits.\n\n## Install\n\n**Let your AI install it.** Tell Claude Code \u002F Codex \u002F OpenClaw \u002F Gemini CLI:\n\n> Install the site-memory skill from github.com\u002FLittleYier\u002Fsite-memory\n\nOr use the [Skills CLI](https:\u002F\u002Fgithub.com\u002Fvercel-labs\u002Fskills):\n\n```bash\nnpx skills add LittleYier\u002Fsite-memory\n```\n\nOr clone manually:\n\n```bash\ngit clone --recurse-submodules https:\u002F\u002Fgithub.com\u002FLittleYier\u002Fsite-memory.git\n```\n\nThen copy `skills\u002Fsite-memory\u002F` to your agent's skills directory\n(e.g. `~\u002F.claude\u002Fskills\u002F`, `~\u002F.gemini\u002Fskills\u002F`, or `.agents\u002Fskills\u002F`).\n\n\u003C!-- AGENT INSTALL INSTRUCTIONS\nIf you are an AI agent installing this skill, follow these steps:\n\n1. Run: npx skills add LittleYier\u002Fsite-memory\n   This handles everything automatically. If npx skills is unavailable, do steps 2-4 instead.\n\n2. git clone --recurse-submodules https:\u002F\u002Fgithub.com\u002FLittleYier\u002Fsite-memory.git \u002Ftmp\u002Fsite-memory\n3. cp -r \u002Ftmp\u002Fsite-memory\u002Fskills\u002Fsite-memory \u003Cskills-dir>\u002Fsite-memory\n   where \u003Cskills-dir> is ~\u002F.gemini\u002Fskills, .agents\u002Fskills, or equivalent.\n4. rm -rf \u002Ftmp\u002Fsite-memory\n\nVerify: test -f \u003Cskills-dir>\u002Fsite-memory\u002FSKILL.md\nRequires: Node.js 22+\nBrowser tool: works with any browser tool the agent has access to. Chrome CDP skill bundled but optional.\nInvoke with: \u002Fsite-memory \u003Ctask description>\nMemory is stored at: ~\u002F.site-memory\u002F (automatic, no config needed)\n-->\n\n## Prerequisites\n\n- Node.js 22+\n\n## Browser tool options\n\nWorks with any browser tool your agent already has:\n\n| Tool | Notes |\n|------|-------|\n| Chrome DevTools Protocol (bundled) | Installed with site-memory, requires Chrome with remote debugging |\n| [browser-use](https:\u002F\u002Fgithub.com\u002Fbrowser-use\u002Fbrowser-use) | `npx skills add anthropics\u002Fbrowser-use` |\n| Playwright MCP | Add the MCP server to your agent config |\n| [Claude in Chrome](https:\u002F\u002Fchromewebstore.google.com\u002Fdetail\u002Fclaude-in-chrome\u002F) | Browser extension, accessed as an MCP tool |\n| Others | Anything that can navigate, read, and interact with pages |\n\nMost agent runtimes (Claude Code, Gemini CLI, Codex) can load multiple skills at once. site-memory works with whatever browser tool you pick.\n\n## Usage\n\nBest for repeated browser work, such as:\n\n```text\nFill out the insurance quote form on example-insurance.com\nSearch for apartments under $300k on realestate-site.com\nGo to the admin dashboard and export last month's report\n```\n\nOr call it directly:\n\n```text\n\u002Fsite-memory open books.toscrape.com and get page 3 titles\n```\n\n## Memory location\n\n`~\u002F.site-memory\u002F`\n\n## Benchmark\n\nTested on [WebVoyager](https:\u002F\u002Fgithub.com\u002FMinorJerry\u002FWebVoyager) tasks.\n**15 websites, 50 tasks, 3 rounds each (150 total runs). All correct.**\n\nRound 1: Claude Sonnet 4.6. Rounds 2 and 3: Claude Haiku 4.5 (3x cheaper).\n\n| Site | R1 avg cmds | R2 avg cmds | R3 avg cmds | R1 avg time | R3 avg time | Cmd delta | Time delta |\n|------|------------:|------------:|------------:|------------:|------------:|----------:|-----------:|\n| Booking | 35.7 | 6.7 | 5.0 | 497s | 30s | **-86%** | **-94%** |\n| Amazon | 14.7 | 2.0 | 2.0 | 183s | 12s | **-86%** | **-93%** |\n| ArXiv | 9.5 | 1.5 | 1.5 | 143s | 9s | **-84%** | **-94%** |\n| Google Flights | 23.3 | 10.3 | 4.0 | 173s | 47s | **-83%** | **-73%** |\n| ESPN | 8.3 | 2.3 | 2.0 | 85s | 18s | **-76%** | **-78%** |\n| Huggingface | 17.7 | 4.3 | 4.3 | 360s | 43s | **-75%** | **-88%** |\n| BBC News | 7.7 | 3.0 | 2.0 | 115s | 22s | **-74%** | **-81%** |\n| Allrecipes | 3.8 | 1.0 | 1.0 | 85s | 3s | **-73%** | **-96%** |\n| Coursera | 3.0 | 2.0 | 1.0 | 45s | 5s | **-67%** | **-89%** |\n| Google Map | 10.7 | 5.3 | 7.0 | 273s | 159s | -34% | -42% |\n| Google Search | 7.5 | 4.0 | 5.3 | 48s | 38s | -29% | -21% |\n| Cambridge Dictionary | 3.0 | 2.3 | 2.3 | 48s | 30s | -23% | -37% |\n| Wolfram Alpha | 4.8 | 5.0 | 3.5 | 45s | 21s | -27% | -53% |\n| GitHub | 2.3 | 2.3 | 2.0 | 33s | 16s | -13% | -52% |\n| Apple | 4.7 | 2.3 | 6.7 | 60s | 72s | +43% | +19% |\n\n### Aggregate results (50 tasks)\n\n| Metric | Round 1 | Round 2 | Round 3 | R1 → R3 |\n|--------|--------:|--------:|--------:|---------:|\n| Total commands | 498 | 177 | 162 | **-67%** |\n| Total time | 6,944s | 1,746s | 1,674s | **-76%** |\n| Avg commands \u002F task | 10.0 | 3.5 | 3.2 | **-68%** |\n| Avg time \u002F task | 139s | 35s | 33s | **-76%** |\n\nAfter just one visit, the second run is already 3x faster.\nFor stable sites (Allrecipes, Coursera, ArXiv), the agent goes straight to the goal after a few visits.\n\n## License\n\nMIT\n","site-memory 是一个用于浏览器自动化AI代理的持久内存工具，能够自动学习网站知识，并在每次访问后更新以不断提高效率。其核心功能包括自动学习站点信息、每次访问后自动更新以及随着时间推移性能逐步提升。该工具特别适用于需要重复执行相同任务的场景，如在线预订、填写表单等，可显著减少操作步骤和时间成本，提高工作效率。基于JavaScript开发，兼容多种浏览器工具，易于集成到现有的AI代理环境中。","2026-06-11 02:43:11","CREATED_QUERY"]