hive-browser

hive-browser is a skill for Claude Code, Codex from dip497/hivemind. It costs 192 tokens per session (2,012 once invoked), scanned A, original, MIT.

A browser-control skill for agents running inside a hivemind workspace. It operates the visible Chromium browser tile that the user is watching, rather than a hidden browser.

In plain words
What is it for?
It helps open websites, navigate pages, click controls, fill out forms, read page content, and capture screenshots in a visible browser tile.
Why use it?
It lets the agent browse in the same page and browser session as the user, so both can see the navigation, clicks, form entries, page content, and screenshots.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/dip497/hivemind/hive-browser
Any agent
npx skills add dip497/hivemind --skill hive-browser
Clone the repo
git clone --depth 1 https://github.com/dip497/hivemind

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for hive-browser

README.md
[![agentmods](https://agentmods.dev/badge/skills/dip497/hivemind/hive-browser.svg)](https://agentmods.dev/skills/dip497/hivemind/hive-browser)
Your own site
<a href="https://agentmods.dev/skills/dip497/hivemind/hive-browser"><img src="https://agentmods.dev/badge/skills/dip497/hivemind/hive-browser.svg" alt="Measured on agentmods" height="20"></a>
Per session 192 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,012 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00192 $0.02012
Opus 5 $0.00096 $0.01006
Sonnet 5 $0.00038 $0.00402
Haiku 4.5 $0.00019 $0.00201

Measured 5d ago against content hash db3e90bf4565, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

hive-browser scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

templates/agentic/.claude/skills/hive-browser/SKILL.md · 176 lines

How it starts

The opening of the file, as written. The whole thing — 176 lines — stays where its author put it; the contents beside it link to each section on GitHub.

hive-browser

You are an agent running inside a hivemind tile (a terminal on an infinite canvas). The canvas can host Browser tiles — real Chromium web views the user can see. This skill lets you drive one of those tiles: same pixels the user is watching, controlled over the Chrome DevTools Protocol (CDP) using the agent-browser CLI.

You are not spawning a hidden headless browser. You attach to the user's visible tile, act in it, and they watch it happen. That shared context is the whole point.

How it works (the mental model)

  • Each Browser tile is an Electron <webview> with its own webContents.
  • hivemind exposes a loopback CDP port (when enabled) and writes a discovery file listing the open browser tiles (tile id, frame, URL).
  • agent-browser --cdp <port> connects to that endpoint; you pick the tab that is the user's tile and drive it with open / snapshot / click / fill / screenshot.

Step 0 — Preconditions (check these first)

Run these and reason about the output before doing anything else:

# Is the CDP bridge enabled, and where is the discovery file?
echo "targets=$HIVEMIND_BROWSER_TARGETS  port=$HIVEMIND_BROWSER_CDP_PORT"
cat "$HIVEMIND_BROWSER_TARGETS" 2>/dev/null || echo "NO_DISCOVERY_FILE"

Interpret:

  • port= is empty / cdpEnabled:false in the file → the CDP bridge is OFF. It is opt-in for safety (a debug port also exposes the app window). Tell the user to enable it: Settings (gear, top-right) → "Enable agent browser control" → Relaunch to apply (or set HIVEMIND_BROWSER_CDP=1 in the environment before launch). Then retry. Do not try to work around this.
  • NO_DISCOVERY_FILE or tiles: [] → no Browser tile is open. Ask the user to open one (canvas hotkey 7, or a frame's + → Browser), ideally in the same frame as your tile. Then re-read the file.
  • tiles: has entries → good, continue.

Make sure the agent-browser CLI is reachable. Prefer a global install if one exists; otherwise run it on demand with npx — it fetches and caches the native binary on first use, so there's no global install and no extra permissions:

Read the full file on GitHub · 176 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 176 lines · 192 tokens per session scan A db3e90bf4565

Subscribe to this mod's changes

hive-browser is a skill published in the GitHub repository dip497/hivemind (6 stars, last pushed yesterday), licensed MIT. It adds 192 tokens to every session and 2,012 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

browser

Browser automation via the agent-browser CLI. Use when the user needs to drive websites or Electron desktop apps — navigating, filling forms, clicking, screenshots, extracting data, testing web apps, visual UI checks, the Pi Dashboard's Electron shell, or the user's own logged-in browser (SSO/2FA sites). Triggers…

BlackBeltTechnology/pi-agent-dashboard · 92 tokens

run-dashboard-e2e-local-changes

Run Playwright E2E (tests/e2e/) against the docker/ all-in-one harness so it reflects LOCAL code changes, not a stale cached image.

BlackBeltTechnology/pi-agent-dashboard · 42 tokens

qa

QA testing skill with real browser automation. Use when asked to "test this site", "QA this page", "check for visual bugs", "verify the deploy", or when Hydra needs browser validation for UI changes. Requires the browse binary.

blueberrycongee/termcanvas · 50 tokens

agency-agentic-search-optimizer

Expert in WebMCP readiness and agentic task completion — audits whether AI agents can actually accomplish tasks on your site (book, buy, register, subscribe), implements WebMCP declarative and imperative patterns, and measures task completion rates across AI browsing agents.

BlackPearl-AI/BlackPearl-CodingAgent · 58 tokens

record-browser-gif

Record browser or Web UI interaction demos as optimized GIFs using the available built-in browser, state-based frame capture, and deterministic encoding, then publish to a dedicated assets branch when the task includes attaching the GIF to a pull request. Use when asked to make, record, or generate a GIF that…

BlackPearl-AI/BlackPearl-CodingAgent · 99 tokens

webapp-testing

Start/reuse a local app, wait for readiness, inspect rendered state/console/network, act from observed selectors, and verify with evidence.

Hmbown/CodeWhale · 32 tokens