analyze-video

Analyze a video or image file using the Qwen Vision Bridge.

Input

• $1 — path to the video or image file
• Remaining arguments ($ARGUMENTS minus $1) — optional analysis prompt. If not provided, use a general analysis prompt appropriate to the file type.

Steps

1. Verify the file exists at the given path. If not, search the current directory and workspace for a matching filename.

2. Check that DASHSCOPE_API_KEY is set:

   echo $DASHSCOPE_API_KEY | head -c 5
   

   If empty, tell the user to run /qwen-setup first.

3. Ensure dashscope is installed:

   python3 -c "import dashscope" 2>&1
   

   If missing, install it: pip install dashscope --break-system-packages

4. Run the bridge:

   python3 "${CLAUDE_PLUGIN_ROOT}/skills/qwen-vision/scripts/qwen_bridge.py" "$1" "prompt here" --fps 2
   

5. Present the analysis to the user in a clean, readable format. If the user's prompt implies structured output (classification, comparison, scoring), format accordingly.

6. Ask if the user wants to:

   • Follow up with more questions about the same video (use --save-context)
   • Analyze another file
   • Save the analysis to a file

For batch analysis

If the user provides a directory path or glob pattern instead of a single file, write a batch script following the pattern in ${CLAUDE_PLUGIN_ROOT}/skills/qwen-vision/references/batch-pattern.md and run it.