Hermes Agent identifies pictures in CLI. The most stable way is to use/pasteorAlt+Vto check the clipboard pictures. OrdinaryCtrl+Vis only suitable for cases where there are both text and pictures on the clipboard; if the clipboard only has pictures, many terminals will not transfer anything.How to handle each platform
pngpaste_.xclip.wl-clipboard.- WSL2: The Windows clipboard can usually be read through
powershell.exe.- VS Code Terminal: Alt key combination may be blocked, use
/pastefirst. - VS Code Terminal: Alt key combination may be blocked, use
Why not SSH remote?SSH Hermes runs on a remote machine, and the clipboard tool reads the remote clipboard, not your local computer. The solution is to upload the image file, give the image URL, or send the image to Hermes via Telegram, Discord, or Slack.
In a word: Use/paste for local terminals, use files or URLs for remote SSH. Don't expect that Ctrl+V will take effect for pure images.
What should I do if the image is not recognized after transferring?First make sure that the model or auxiliary vision model you configured really supports vision. Then check whether the terminal can read the clipboard: Linux looks atxcliporwl-paste, WSL2 looks atpowershell.exe, and remote SSH uses the file path or picture URL instead. Don't repeat the blind test of Ctrl+V, first confirm clearly whether the picture has entered Hermes.
Official open source address: https://github.com/NousResearch/hermes-agent; Official document entry: https://hermes-agent.nousresearch.com/.
