🛩️ dsh-pilot — give your DSH agent hands

Drive a real browser from the DeepSeek Harness chat: the agent opens pages, reads them as structured text with a numbered element list, clicks and types by ref (no CSS guessing), presses keys, navigates back/reload, waits, evaluates JS, and takes screenshots — while you watch a live draggable cockpit panel in the Web GUI and can take over at any time.

• 🚀 One command install — dsh plugin --profile web add github:guo6x/dsh-pilot
• ⚡ Zero runtime dependencies — talks CDP over the native Node ≥ 22 WebSocket, uses the Edge/Chrome already on your machine
• 🔑 No API key — nothing leaves your machine; no vision model required
• 📖 Text-first by design — the agent reads DOM snapshots (title/URL/text/links + numbered elements), so text-only models browse without burning vision tokens
• 🎯 Ref-driven interaction — every click/type targets a snapshot ref, not a guessed selector; stale refs fail loudly with a hint
• 📝 Human-named forms — fill several fields by label, aria-label, placeholder, name, or id, and upload files by ref
• 🧭 Full navigation set — back, reload, and wait tools for real browsing flows, with page-settling waits built in
• 👀 Human in the loop — live screenshot, URL bar, action log, and a session switcher in the cockpit; you see everything the agent does
• 🧩 Per-session isolation — every agent session gets its own browser instance; parallel sessions never fight over one page

Watch it work

Open the ✈️ cockpit, start its isolated browser, and watch the page change live. When several agents are browsing, the demo also pins one session, then returns to automatic following.

Install — copy, paste, confirm

# Install from GitHub — this is the supported release channel.
dsh plugin --profile web add github:guo6x/dsh-pilot

Restart a running dsh web process, then refresh the page. Installation is complete when a ✈️ button appears at the bottom of the sidebar. Click it to open the cockpit.

Requirements: the DeepSeek Harness web profile, Node ≥ 22, and Edge or Chrome installed. The plugin needs no extra account, API key, or browser download.

Developing from a checkout instead? Run dsh plugin --profile web add . from the repository directory. The committed lib/ files mean GitHub installs do not run a build script.

First safe task

Start a new chat and paste this:

Open https://example.com, tell me what the page says, then verify that the “Learn more” link is visible.

You should see a structured answer with the page title and a numbered link, while the cockpit shows the same page. This task only reads a public page.

If the ✈️ button is missing

• Confirm the plugin is installed in the web profile: dsh plugin --profile web list dsh-pilot.
• Restart the dsh web process after installing; a browser refresh alone cannot load new host code.
• Check that Node is version 22 or newer and that Edge or Chrome is installed; the cockpit reports a launch error when neither is available.

See it work in 60 seconds

For a real staging form, give the agent one bounded task:

Open our staging application form. Fill the contact fields, upload the approved resume at C:\\path\\to\\resume.pdf, verify the entered values, and stop before final submission.

…