Screenshot Telegram Direct

Capture website screenshots and send to Telegram via direct API. Works around OpenClaw's broken image delivery (issue

Security Scan

The skill's code and docs perform the claimed screenshot→Telegram task, but registry metadata omits the three required secrets and it relies on an external snap API (snap.llm.kaveenk.com); those inconsistencies and external dependencies warrant caution.

Purpose & Capability

The script and SKILL.md both implement capturing a website via an external screenshot API and posting to Telegram — this matches the advertised purpose. However the registry metadata lists no required environment variables, while both SKILL.md and screenshot-send.sh require three secrets (TELEGRAM_BOT_TOKEN, TELEGRAM_CHAT_ID, SNAP_API_KEY). That mismatch between declared requirements and actual runtime needs is a material incoherence.

Instruction Scope

Runtime instructions and the script only call the snapshot service (snap.llm.kaveenk.com) and Telegram API (api.telegram.org), and write a temporary file in /tmp which is removed on success. These actions are within the stated purpose. The docs do recommend auto-sourcing a .env from the skill directory into the user's shell profile (~/.bashrc / ~/.zshrc), which broadens scope by making secrets available to every shell session — this is a potentially risky recommendation and should be optional and clearly explained.

Install Mechanism

No install spec; this is an instruction-only skill with a single helper script. Installation is limited to copying .env, making the script executable, and optionally adding a cron entry — nothing is downloaded or written by an automated installer.

Credentials

The skill requires three sensitive environment values (TELEGRAM_BOT_TOKEN, TELEGRAM_CHAT_ID, SNAP_API_KEY) but the registry metadata advertises none. SNAP_API_KEY is sent to an external domain (snap.llm.kaveenk.com) — trust in that service is required because it will receive the URL and any content necessary to produce screenshots. Recommending global auto-sourcing of .env increases the blast radius if those secrets are compromised.

Persistence & Privilege

The skill does not request always:true and does not modify other skills or system settings. The only persistent change the docs suggest is editing a shell profile to auto-load the .env, which is a user action (not automatic) but exposes secrets to all shell sessions if applied.

What to consider before installing

Key points before installing: (1) Metadata inconsistency — the registry says no env vars but the script requires TELEGRAM_BOT_TOKEN, TELEGRAM_CHAT_ID, and SNAP_API_KEY. Treat that as a red flag and verify with the author. (2) The script sends the SNAP_API_KEY and the target URL to snap.llm.kaveenk.com; confirm you trust that third party (or run your own screenshot service). (3) Avoid automatically sourcing .env from ~/.bashrc unless you understand the risk of exposing tokens to all shells; prefer sourcing manually per session or using a constrained wrapper. (4) Keep .env out of version control, rotate tokens after testing, and test with a throwaway bot/chat and a non-sensitive URL. (5) If you need higher assurance, ask the publisher for a homepage or source repository, or replace the external snap endpoint with a known/trusted service or self-hosted screenshot tool.

Capture website screenshots and send them to Telegram, bypassing OpenClaw's broken image delivery pipeline.

Quick Start

# 1. Setup environment
cp .env.example .env
# Edit .env with your tokens

# 2. Run the helper script
./screenshot-send.sh https://github.com

Setup

1. Get API Credentials

Telegram Bot Token:

• Message @BotFather
• Send /mybots → Select bot → Copy API token

Telegram Chat ID:

• Personal: Message @userinfobot
• Group: Add @RawDataBot to group

Snap API Key:

# Register for free API key
curl -s -X POST https://snap.llm.kaveenk.com/api/register \
  -H "Content-Type: application/json" \
  -d '{"name":"my-agent"}'
# Save the "key" from response

2. Configure

cp .env.example .env
# Edit .env:
# TELEGRAM_BOT_TOKEN=your_token
# TELEGRAM_CHAT_ID=your_chat_id
# SNAP_API_KEY=your_snap_key

# Source it (temporary)
source .env

3. Auto-Load .env (Recommended for OpenClaw)

Option A: Add to shell profile (always available)

Add to ~/.bashrc or ~/.zshrc:

# Auto-load screenshot-telegram-direct env vars
SKILL_DIR="$HOME/.openclaw/workspace/skills/screenshot-telegram-direct"
if [ -f "$SKILL_DIR/.env" ]; then
    export $(grep -v '^#' "$SKILL_DIR/.env" | xargs)
fi

Then reload: source ~/.bashrc

Option B: Source before each OpenClaw session

In your OpenClaw workspace, create load-env.sh:

#!/bin/bash
source "$HOME/.openclaw/workspace/skills/screenshot-telegram-direct/.env"
echo "✅ Environment loaded from screenshot-telegram-direct/.env"

Run before using: source load-env.sh

Option C: Modify screenshot-send.sh to auto-source

The script already checks for env vars. To auto-source .env if it exists, add at the top:

# Auto-source .env if it exists in script directory
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
if [ -f "$SCRIPT_DIR/.env" ]; then
    source "$SCRIPT_DIR/.env"
fi

3. Make Script Executable

chmod +x screenshot-send.sh

Usage

Basic screenshot

./screenshot-send.sh https://example.com

Custom caption

./screenshot-send.sh https://github.com "Latest PR status"

Multiple URLs

for url in "https://github.com" "https://news.ycombinator.com"; do
  ./screenshot-send.sh "$url"
  sleep 2
done

Configuration Options

Snap API Options

Edit screenshot-send.sh to customize:

# Full page screenshot
curl -s -X POST "https://snap.llm.kaveenk.com/api/screenshot" \
  -H "Authorization: Bearer $SNAP_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "'$URL'",
    "full_page": true,
    "width": 1920,
    "height": 1080,
    "dark_mode": true,
    "wait_ms": 2000
  }' \
  -o "$OUTPUT_FILE"

Option	Description
full_page	Capture entire scrollable page
width/height	Viewport size
dark_mode	Emulate dark color scheme
wait_ms	Wait for JS rendering
format	png or jpeg

Send as Document (No Compression)

Replace sendPhoto with sendDocument in the script:

curl -s -X POST \
  -F "chat_id=$CHAT_ID" \
  -F "document=@$OUTPUT_FILE" \
  -F "caption=$CAPTION" \
  "https://api.telegram.org/bot$TOKEN/sendDocument"

Automation (Cron)

# Daily dashboard screenshot
0 9 * * * cd /home/pi/.openclaw/workspace/skills/screenshot-telegram-direct && ./screenshot-send.sh https://status.example.com

# Hourly monitoring
0 * * * * cd /home/pi/.openclaw/workspace/skills/screenshot-telegram-direct && ./screenshot-send.sh https://uptime.example.com

File Structure

screenshot-telegram-direct/
├── SKILL.md              # This documentation
├── .env.example          # Template for env vars
├── .env                  # Your secrets (gitignored)
├── .gitignore            # Excludes .env
└── screenshot-send.sh    # Main script

Security

• ✅ .env is gitignored — never commit secrets
• ✅ .env.example shows required variables without values
• ✅ Script fails if env vars are missing
• ✅ No hardcoded credentials

Troubleshooting

Issue	Fix
Missing environment variables	Check .env exists and is sourced
Failed to capture screenshot	Check SNAP_API_KEY is valid
chat not found	Verify TELEGRAM_CHAT_ID
bot was blocked	Send /start to bot in Telegram

Why Not Use OpenClaw Native?

See OpenClaw issue #63137: Images sent via read tool render locally but never reach mobile clients — silent failure with no error logs.

This skill bypasses OpenClaw's Telegram channel entirely using direct HTTP API calls.

Publishing Notes

This skill is ready for publication:

• No hardcoded secrets
• .env.example for configuration
• .gitignore excludes secrets
• Clear documentation
• MIT licensed