alvin-bot 4.4.1

package/skills/email-summary/SKILL.md

@@ -0,0 +1,46 @@
+---
+name: Email Summary
+description: Create professional email summaries and daily digests
+triggers: email summary, email digest, mail zusammenfassung, mails zusammenfassen, email übersicht, inbox summary, email report
+priority: 5
+category: productivity
+---
+
+# Email Summary Skill
+
+When creating email summaries or daily digests:
+
+## Workflow
+1. **Fetch emails** using available tools (osascript for Apple Mail, himalaya for IMAP)
+2. **Summarize YOURSELF** — you ARE the AI, do NOT call external LLM APIs
+3. **Structure** the output clearly (by account, by priority, by topic)
+4. **Deliver** in the requested format (text, PDF, or both)
+
+## Summary Guidelines
+- **1-2 sentences per email** — capture the core message, skip fluff
+- **Language:** Match the user's preferred language
+- **Prioritize:** Flag urgent/action-required emails prominently
+- **Group by:** Account → Read status (unread first) → Time
+
+## Apple Mail Access (macOS)
+```bash
+osascript -e '
+tell application "Mail"
+  set output to ""
+  repeat with acct in accounts
+    set acctName to name of acct
+    repeat with mb in mailboxes of acct
+      if name of mb is "INBOX" then
+        set msgs to (messages of mb whose date received > (current date) - 1 * days)
+        repeat with m in msgs
+          set output to output & acctName & " | " & sender of m & " | " & subject of m & " | " & (date received of m as string) & " | " & (read status of m as string) & linefeed
+        end repeat
+      end if
+    end repeat
+  end repeat
+  return output
+end tell'
+```
+
+## PDF Generation
+If a PDF is requested, create an HTML file and convert it. Use a clean, professional design.

package/skills/github/SKILL.md

@@ -0,0 +1,42 @@
+---
+name: GitHub
+description: Manage GitHub issues, PRs, and repos via gh CLI
+triggers: github, gh, issue, pull request, pr, repository, repo, merge, branch
+priority: 6
+category: development
+---
+
+# GitHub via gh CLI
+
+Use the GitHub CLI (`gh`) for all GitHub operations.
+
+## Issues
+```bash
+gh issue list
+gh issue create --title "Title" --body "Body"
+gh issue view 123
+gh issue close 123
+```
+
+## Pull Requests
+```bash
+gh pr list
+gh pr create --title "Title" --body "Body"
+gh pr view 123
+gh pr merge 123
+gh pr checks 123
+```
+
+## Repository
+```bash
+gh repo view
+gh repo clone owner/repo
+gh release list
+gh release create v1.0.0
+```
+
+## Search
+```bash
+gh search issues "query" --repo owner/repo
+gh search prs "query"
+```

package/skills/summarize/SKILL.md

@@ -0,0 +1,28 @@
+---
+name: Summarize
+description: Summarize URLs, documents, PDFs, and long text
+triggers: summarize, zusammenfassen, summary, zusammenfassung, tldr, tl;dr, digest
+priority: 7
+category: research
+---
+
+# Summarization Workflow
+
+## URL Summarization
+1. Fetch content: `curl -sL URL | head -c 50000`
+2. Summarize the text directly (YOU are the LLM — no external API needed)
+3. Present: Key points, main arguments, conclusion
+
+## PDF Summarization
+1. Extract text: `pdftotext file.pdf -` or `python3 -c "..."`
+2. Summarize directly
+3. If too long: chunk into sections, summarize each, then meta-summarize
+
+## Long Text
+1. If under 10,000 chars: summarize directly
+2. If over 10,000 chars: split into logical sections, summarize each, combine
+
+## Output Format
+- **One-line summary** (1 sentence)
+- **Key Points** (3-5 bullets)
+- **Details** (if requested, 2-3 paragraphs)

package/skills/system-admin/SKILL.md

@@ -0,0 +1,39 @@
+---
+name: System Administration
+description: Server management, deployment, monitoring
+triggers: server, deploy, deployment, pm2, nginx, ssh, vps, docker, systemctl, service, logs, monitor, disk space, cpu, memory usage
+priority: 3
+category: devops
+---
+
+# System Administration Skill
+
+When managing servers and deployments:
+
+## Health Check Pattern
+```bash
+# System overview
+uptime && free -h && df -h / && top -bn1 | head -5
+
+# PM2 processes
+pm2 list && pm2 logs --lines 10 --nostream
+
+# Nginx
+nginx -t && systemctl status nginx
+
+# Docker
+docker ps && docker stats --no-stream
+```
+
+## Deployment Checklist
+1. **Build** locally first — never deploy broken code
+2. **Backup** config files before changing them
+3. **Test** after deploy — check logs, hit endpoints
+4. **Rollback plan** — know how to undo your changes
+
+## Guidelines
+- **Always check before changing** — read config files before editing
+- **Backup first** — `cp file file.bak-$(date +%Y%m%d-%H%M%S)`
+- **PM2 env vars** — use `ecosystem.config.cjs` + `--update-env`, never raw `pm2 start`
+- **Logs first** — when debugging, always check logs before guessing
+- **One change at a time** — don't change 5 things and wonder which broke it

package/skills/weather/SKILL.md

@@ -0,0 +1,34 @@
+---
+name: Weather
+description: Get weather forecasts and conditions
+triggers: weather, wetter, forecast, vorhersage, temperature, temperatur, rain, regen, wind
+priority: 4
+category: utility
+---
+
+# Weather
+
+## Quick Weather
+```bash
+curl -s "wttr.in/Berlin?format=%l:+%c+%t+%w+%h"
+```
+
+## Detailed Forecast
+```bash
+curl -s "wttr.in/Berlin?lang=de"
+```
+
+## Specific Location
+```bash
+curl -s "wttr.in/LOCATION?format=3"
+```
+
+## Machine-Readable
+```bash
+curl -s "wttr.in/Berlin?format=j1" | python3 -c "import sys,json; d=json.load(sys.stdin); print(json.dumps(d['current_condition'][0], indent=2))"
+```
+
+## Notes
+- Default city: Berlin (unless user specifies another)
+- Language detection: use `?lang=de` for German users
+- For multi-day: use `wttr.in/CITY?3` for 3-day forecast

package/skills/web-research/SKILL.md

@@ -0,0 +1,35 @@
+---
+name: Web Research
+description: Deep web research with source aggregation and summarization
+triggers: research, recherche, find information, look up, search for, herausfinden, web search, deep dive
+priority: 3
+category: research
+---
+
+# Web Research Skill
+
+When conducting web research:
+
+## Workflow
+1. **Search** — use curl/web_fetch to search and fetch relevant pages
+2. **Read & extract** — parse the content, extract key information
+3. **Synthesize YOURSELF** — combine findings into a coherent summary
+4. **Cite sources** — always include URLs for claims
+
+## Search Pattern
+```bash
+# Google via web scraping
+curl -sL "https://www.google.com/search?q=QUERY&hl=en" | python3 -c "
+import sys, re
+html = sys.stdin.read()
+for m in re.finditer(r'<a href=\"/url\?q=(https?://[^&\"]+)', html):
+    print(m.group(1))
+" | head -5
+```
+
+## Guidelines
+- **Multiple sources** — never rely on a single source
+- **Recency** — prefer recent results for time-sensitive topics
+- **Summarize** — don't dump raw page content, distill the key findings
+- **Structure** — organize by subtopic or relevance
+- **Bias check** — note if sources are promotional or one-sided

package/web/public/canvas.html

@@ -0,0 +1,52 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+  <meta charset="UTF-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1.0">
+  <title>Alvin Bot — Canvas</title>
+  <style>
+    * { margin: 0; padding: 0; box-sizing: border-box; }
+    body { font-family: -apple-system, BlinkMacSystemFont, sans-serif; background: #0a0a0a; color: #e0e0e0; height: 100vh; display: flex; flex-direction: column; }
+    #toolbar { background: #1a1a1a; padding: 8px 16px; display: flex; gap: 12px; align-items: center; border-bottom: 1px solid #333; }
+    #toolbar .status { font-size: 12px; color: #888; margin-left: auto; }
+    #toolbar .status.connected { color: #4ade80; }
+    #canvas-container { flex: 1; position: relative; overflow: auto; }
+    #render-frame { width: 100%; height: 100%; border: none; background: white; }
+    .empty-state { display: flex; align-items: center; justify-content: center; height: 100%; color: #555; font-size: 18px; }
+  </style>
+</head>
+<body>
+  <div id="toolbar">
+    <strong>Canvas</strong>
+    <span class="status" id="status">Disconnected</span>
+  </div>
+  <div id="canvas-container">
+    <div class="empty-state" id="empty">Waiting for content from Alvin Bot...</div>
+    <iframe id="render-frame" style="display:none" sandbox="allow-scripts allow-same-origin"></iframe>
+  </div>
+  <script>
+    const ws = new WebSocket(`ws://${location.host}/canvas/ws`);
+    const status = document.getElementById("status");
+    const empty = document.getElementById("empty");
+    const frame = document.getElementById("render-frame");
+
+    ws.onopen = () => { status.textContent = "Connected"; status.className = "status connected"; };
+    ws.onclose = () => { status.textContent = "Disconnected"; status.className = "status"; };
+
+    ws.onmessage = (event) => {
+      const msg = JSON.parse(event.data);
+      if (msg.type === "present") {
+        empty.style.display = "none";
+        frame.style.display = "block";
+        frame.srcdoc = msg.html;
+      } else if (msg.type === "eval") {
+        try { frame.contentWindow.eval(msg.js); } catch (e) { console.error("Canvas eval error:", e); }
+      } else if (msg.type === "clear") {
+        frame.style.display = "none";
+        empty.style.display = "flex";
+        frame.srcdoc = "";
+      }
+    };
+  </script>
+</body>
+</html>