chevereto-upload

name	chevereto-upload
description	Upload images to a Chevereto V4 image hosting service using a local config file for permanent, shareable URLs. Use when images (AI-generated, screenshots, or any visual content) need to be hosted, archived, or shared via a link. Also supports listing and deleting uploaded images.
metadata	{"version":"0.4.0"}

Chevereto Upload

Upload images to a Chevereto V4 instance via the bundled shell script. The script handles API authentication, file upload, metadata tagging, and upload logging.

How to Upload

Run the upload script with the image path and metadata:

bash <skill_dir>/scripts/upload.sh <file_path> "<title>" "<description>" "<tags>" [album_id]

Provide title, description, and tags on every upload — these make images discoverable and traceable.

Example: AI-Generated Image

bash <skill_dir>/scripts/upload.sh /tmp/2026-03-01-14-30-00-sunset.png \
  "Sunset Over Cyberpunk City" \
  "Prompt: A dramatic sunset over a neon-lit cyberpunk cityscape with flying cars
Model: gemini-3.1-flash-image
Generated: 2026-03-01
Request: cyberpunk sunset wallpaper" \
  "ai-generated,openclaw,gemini,cyberpunk,sunset"

Example: Screenshot or Other Image

bash <skill_dir>/scripts/upload.sh /tmp/screenshot.png \
  "Dashboard Overview Q1" \
  "Screenshot of the analytics dashboard showing Q1 2026 metrics" \
  "screenshot,dashboard,analytics"

Metadata Guide

Field	How to Fill
title	Short descriptive name in English (e.g., "Cherry Blossom Phone Wallpaper")
description	For AI images: include Prompt, Model, Generated date, and user's original Request. For other images: describe content and context
tags	Comma-separated keywords. AI images always include ai-generated,openclaw plus model name and style tags

Metadata rules:

• Never include proxy/third-party service info in descriptions — only the model name (e.g., "gemini-3-pro-image", not "via gptclub proxy")
• Always upload via the script — do not use raw curl against the API. The script ensures all metadata fields (delete_url, dimensions, size) are logged to chevereto-uploads.jsonl
• Non-ASCII titles (Chinese, etc.) are handled gracefully — the script falls back to a timestamp-based filename when the title has no ASCII-safe characters

Response

The script outputs JSON with:

• viewer_url — shareable page link (browsable, shows metadata)
• direct_url — direct image file URL (for embedding)
• thumb_url — thumbnail URL
• delete_url — one-time deletion link

Return both viewer_url and direct_url to the user after upload.

All uploads are automatically logged to memory/chevereto-uploads.jsonl.

List Uploads

bash <skill_dir>/scripts/list.sh                     # all uploads
bash <skill_dir>/scripts/list.sh --recent 10          # last 10
bash <skill_dir>/scripts/list.sh --search "keyword"   # search by title

Delete

bash <skill_dir>/scripts/delete.sh <image_id_or_delete_url>

Accepts an image ID (looks up delete_url from the upload log) or a full delete URL.

Configuration

The upload script reads a local JSON config file. Default path:

/root/.openclaw/config/chevereto-upload.json

Optional override: set CHEVERETO_CONFIG to another config file path for one command. Do not store the API key in OpenClaw env.vars.

{
  "url": "https://imglab.cc",
  "api_key": "<chevereto api key>",
  "album_id": "VBQ",
  "log_file": "/root/.openclaw/workspace/memory/chevereto-uploads.jsonl"
}

Field	Required	Description
api_key	Yes	User API key
url	No	Site URL (default: https://imglab.cc)
album_id	No	Default album ID for uploads
log_file	No	Upload log path