🤖 KI-Agenten & Workflows

GitHub Stars Fetcher with Agent Browser

📁 KI-Agenten & Workflows 👤 Beigetragen von @XiaoStore 🗓️ Aktualisiert
Der Prompt
# Using Agent Browser to Fetch GitHub Starred Projects ## Objective Use the Agent Browser skill to log into GitHub and retrieve the starred projects of the currently logged-in user, sorted by the number of stars. ## Execution Steps (Follow in Order) 1. **Launch Browser and Open GitHub Homepage** ```bash agent-browser --headed --profile "%HOMEPATH%\.agent-browser\chrome-win64\chrome-profiles\github" open https://github.com && agent-browser wait --load networkidle ``` 2. **Get Current Logged-in User Information** ```bash agent-browser snapshot -i # Find the user avatar or username link in the top-right corner to confirm login status # Extract the username of the currently logged-in user from the page ``` 3. **Navigate to Current User's Stars Tab** ```bash # Construct URL: https://github.com/{username}?tab=stars agent-browser open https://github.com/{username}?tab=stars && agent-browser wait --load networkidle ``` 4. **Sort by Stars Count (Most Stars First)** ```bash agent-browser snapshot -i # First get the latest snapshot to find the sort button agent-browser click @e_sort_button # Click the sort button agent-browser wait --load networkidle # Select "Most stars" from the dropdown options ``` 5. **Retrieve and Record Project Information** ```bash agent-browser snapshot -i # Extract project name, description, stars, and forks information ``` ## Critical Notes ### 1. Daemon Process Issues - If you see "daemon already running", the browser is already running - **Important:** When the daemon is already running, `--headed` and `--profile` parameters are ignored, and the browser continues in its current running mode - You can proceed with subsequent commands without reopening - To restart in headed mode, you must first execute: `agent-browser close`, then use the `--headed` parameter to reopen ### 2. Dynamic Nature of References - Element references (@e1, @e2, etc.) change after each page modification - You must execute `snapshot -i` before each interaction to get the latest references - Never assume references are fixed ### 3. Command Execution Pattern - Use `&&` to chain multiple commands, avoiding repeated process launches - Wait for page load after each command: `wait --load networkidle` ### 4. Login Status - Use the `--profile` parameter to specify a profile directory, maintaining login state - If login expires, manually log in once to save the state ### 5. Windows Environment Variable Expansion - **Important:** On Windows, environment variables like `%HOMEPATH%` must be expanded to actual paths before use - **Incorrect:** `agent-browser --profile "%HOMEPATH%\.agent-browser\chrome-win64\chrome-profiles\github"` - **Correct:** First execute `echo $HOME` to get the actual path, then use the expanded path ```bash # Get HOME path (e.g., /c/Users/xxx) echo $HOME # Use the expanded absolute path agent-browser --profile "/c/Users/xxx/.agent-browser/chrome-win64/chrome-profiles/github" --headed open https://github.com ``` - Without expanding environment variables, you'll encounter connection errors (e.g., `os error 10060`) ### 6. Sorting Configuration - Click the "Sort by: Recently starred" button (typically reference e44) - Select the "Most stars" option - Retrieve page content again ## Troubleshooting Common Issues | Issue | Solution | |-------|----------| | daemon already running | Execute subsequent commands directly, or close then reopen | | Invalid element reference | Execute snapshot -i to get latest references | | Page not fully loaded | Add wait --load networkidle | | Need to re-login | Use --headed mode to manually login once and save state | | Sorting not applied | Confirm you clicked the correct sorting option | ## Result Output Format - Project name and link - Stars count (sorted in descending order) - Forks count - Project description (if available)

So nutzt du diesen Prompt

Kopiere den Prompt oben oder klicke einen "Öffnen in"-Button um ihn direkt in deiner bevorzugten KI zu starten. Du kannst den Text dann an deinen Anwendungsfall anpassen — z.B. Platzhalter wie [dein Thema] durch echten Kontext ersetzen.

Welches KI-Modell funktioniert am besten

Claude glänzt bei Agent-Workflows dank langem Context-Window (bis 1M Tokens) und nuancierter Instruction-Following. ChatGPT hat native Actions (Tool-Calling) eingebaut. Gemini integriert am besten mit Google Workspace. Für autonome Workflows ist Claude Sonnet 4.6 aktueller Sweet-Spot für Qualität und Kosten.

Diesen Prompt anpassen

Passe Rolle und Constraints des Agents an deine Umgebung an. Wenn der Prompt bestimmte Tools erwähnt (Search, File I/O, Code-Execution), entferne was du nicht hast und ergänze was du brauchst. Füge Guardrails hinzu: "Immer Bestätigung einholen bevor Dateien geschrieben werden." Definiere Erfolgskriterien explizit.

Typische Anwendungsfälle

  • Autonome Forschungs-Assistenten für einen Bereich bauen
  • Chatbots mit definierten Persönlichkeiten + Wissensgrenzen erstellen
  • Multi-Step-Workflows orchestrieren (Recherche → Entwurf → Review → Publish)
  • System-Prompts für Custom GPTs oder Claude Projects definieren
  • Agent-Loops bauen die Tools rufen und sich selbst korrigieren

Variationen

Passe den Tonfall an (lockerer, technischer), ändere das Ausgabeformat (Aufzählungen vs. Absätze) oder füge Einschränkungen hinzu (Wortlimits, Zielgruppe).

Verwandte Prompts