Menu
Launch and control a local headless Chrome instance for screenshots, PDFs, and debugging. Uses the system's /usr/bin/google-chrome in headless mode.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Thin shell wrapper over the Linear GraphQL API for use from agent directives. Reads Linear projects + issues, creates/updates issues, resolves viewers/cycles/labels/project IDs. Replaces the legacy src/linear/*.ts; orchestration logic now lives in vault directives (linear-import.md, linear-uplink.md), this skill only owns the HTTP boundary.
Multi-reference image composition via fal-ai/nano-banana/edit. Use when a directive needs to combine 2+ reference images into a single composed output (character + location, brand sigil + character, multi-character ensemble shot, etc.). Sister tool: bible-to-prompt assembles a canonical prompt from a character/location bible MD before calling nano-compose.
Fire a single openclaw notification per runtime contract (notifications). Always echoes to stdout in §15.4 format; also dispatches via openclaw CLI if OPENCLAW_NOTIFY_TARGET is set. Replaces src/notify/notify.ts (99 LOC); batching responsibility shifts to callers.
Fetch project-scoped context from Notion using the official REST API for use in <workplace>ProjectSync and other orchestrators.
Read, write, and search Obsidian vault markdown files
Spawn a Claude Code or Cursor agent in non-interactive mode (`--print` style) with a prompt + repo workspace. Replaces src/agents/{claudeCodeSpawn,cursorSpawn,spawnAgent}.ts (179 LOC).
| name | headless-browser |
| description | Launch and control a local headless Chrome instance for screenshots, PDFs, and debugging. Uses the system's /usr/bin/google-chrome in headless mode. |
This skill wraps a simple headless Chrome launcher on this machine using the script:
~/clawd/scripts/headless-browser.shChrome binary:
/usr/bin/google-chromeUse this when you want a long-lived headless browser for manual or external control.
Command executed:
cd ~/clawd && ./scripts/headless-browser.sh start
Behavior:
--headless=new--remote-debugging-port=9222--disable-gpu--user-data-dir=/tmp/headless-chrome-profile--no-first-run --no-default-browser-checkWhen the user asks to "launch headless Chrome" or "start a headless browser", run this command and report:
9222Command executed:
cd ~/clawd && ./scripts/headless-browser.sh screenshot <url> <out.png>
Defaults:
<out.png> is omitted, the script uses screenshot.png in the current directory.Use this when the user says things like:
After running, confirm the output file path, e.g.:
Screenshot saved to ~/clawd/screenshot.pngCommand executed:
cd ~/clawd && ./scripts/headless-browser.sh pdf <url> <out.pdf>
Defaults:
<out.pdf> is omitted, the script uses page.pdf in the current directory.Use this when the user asks:
After running, confirm the output file path.
Trigger this skill when the user says things like:
start instances simultaneously; if necessary, ask the user before launching another Chrome PID.headless-browser.sh subcommand from ~/clawd.start: confirmation + debug portscreenshot/pdf: confirmation + output path/usr/bin/google-chrome installation.