An assistant, or a library of browser tools?
Claude in Chrome is Claude’s assistant on paid plans. Customaise works with the agent you choose, on any plan, and keeps the tools it builds.
By Customaise · Reviewed
Compares Claude in Chrome’s current documented release and permissions with Customaise 1.3.9. Historical security evaluations are not treated as present-day risk rates.
Any agent: Your choice of agent
Claude Code, Cursor, Codex, Kiro, Antigravity, Claude Desktop or Grok Bot, on the free plan too.
Build: Your installed script library
Customaise keeps page controls and named agent tools available for future tasks.
Review: Different approval controls
Both require attention to access, instructions and outcomes; neither is a universal safety guarantee.
Side-by-side comparison
| Capability | Customaise | Claude in Chrome |
|---|---|---|
| Main workflow [4] [1] | Create, install, test and reuse scripts with an external agent through MCP or CLI. | Ask Claude to work on browser tasks through Anthropic’s extension and integrations. |
| Repeatable actions [4] [8] | Install an AgentScript with named tools and explicit arguments. Test it as the site changes. | Shortcuts, which can run on a schedule; recorded workflows only in the classic side panel. |
| Current browser session [4] [1] [2] | Uses the connected Chrome profile and the pages you authorise. | Works through its Chrome extension with browser access under Claude’s controls. |
| Page customisation [3] [1] | UserScripts can add persistent controls, filters and presentation changes on matching pages. | Its primary interface is Claude’s browser assistant; custom script-library management is a different workflow. |
| Approvals [4] [2] | Allow, Prompt and Deny combine with selected approval modes and saved decisions. A script an agent creates over MCP or the CLI cannot grant its own tools Allow, and an agent cannot switch on Full access. | Manually approve, Automatically approve (the Cowork side panel’s default, in which Claude reviews each action for safety) and Skip all approvals. Purchases and account creation are blocked in every mode. |
| Prompt is not a promise of a fresh modal for every Customaise call. Full access changes eligible calls. | ||
| Agent choice [4] [6] [1] [10] | Claude Code, Cursor, Codex, Kiro, Antigravity and Claude Desktop over MCP; Grok Bot and any shell agent over the CLI. Power User can also run the in-browser AI on your own key or your own model server (llama.cpp, Ollama, LM Studio, vLLM). | Uses Claude and the access included in the user’s eligible Claude account. |
| AI decisions inside scripts [5] [1] [10] | Free on-device AI with CM_promptAI (Gemini Nano, no key, Chrome 148 or later). Power User adds Script Decisions with Jev, OpenAI Decisions API or Gemini Decisions API. | Claude performs model reasoning within its browser workflow; that is separate from Customaise Script Decisions. |
| Data and privacy [7] [1] [2] | Local scripts plus the external agent and any enabled cloud features. | Browser context used for Claude tasks is handled under Anthropic’s product and privacy controls. |
| Sharing saved work [4] [6] [1] | Power User can publish scripts or Send an independent copy for another user to enable. | Workflow sharing depends on the Claude feature and account; no Customaise script subscription is implied. |
| Cost and limits [6] [1] [9] | Free bridge allowance, optional Power User and separate agent/provider usage. | Paid Claude plans only (Pro, Max, Team, Enterprise); not the Free plan. Browser tasks count against the plan’s usage limits. |
| Claude Code may itself use a Claude subscription. Using an IDE agent does not necessarily move work off your Claude plan. | ||
What is the real difference?
Claude in Chrome is generally available to paid Claude users according to Anthropic’s August 2026 announcement. It combines browser interaction with Claude’s wider workflow. Customaise is an extension and bridge that a compatible agent can use to write and call installed scripts.
These can be complementary choices. A Claude Code setup can use Customaise’s MCP bridge; that does not mean Claude in Chrome automatically discovers every Customaise tool.
Compare permissions in their actual modes
Declared tools use Allow, Prompt or Deny. The selected approval mode and saved decisions affect whether a Prompt tool asks again; Full access permits eligible calls. Review source and permissions: tool-call controls do not isolate arbitrary script code or an agent with access to the whole computer. [4]
Security evidence needs scope
A red-team result or a report about an older extension version is not a current failure probability. This page does not claim either product is immune to prompt injection or malicious scripts. Review the specific permissions and data you expose. [2] [4] [7]
AI in your scripts
CM_promptAI runs Gemini Nano on the device, free and with no key, in Chrome 148 or later. Power User adds Script Decisions: focused classification or scoring with Jev, OpenAI Decisions API or Gemini Decisions API, declared in the script with @grant CM_decisions and run with a saved provider key on an authorised device. Provider charges are separate. [5] [10]
Usage and pricing
Free includes 50 successful Agent Bridge calls per UTC day and 150 per rolling seven days. Power User removes these bridge caps. Your external agent’s subscription or API charges are separate; Script Decisions provider charges are separate too. [6]
When should you use which?
Customaise
You want reusable page controls and named browser tools, and you are prepared to test and maintain scripts with your chosen agent.
Claude in Chrome
You want integrated Claude browser assistance without first maintaining a separate library of Customaise scripts.
Claude Code can use Customaise through MCP. Configure and verify that connection separately from Claude in Chrome.
Sources and scope
Compares Claude in Chrome’s current documented release and permissions with Customaise 1.3.9. Historical security evaluations are not treated as present-day risk rates.
By Customaise. Product capabilities are based on documentation and our implementation. Recommendations are workflow judgments, not a head-to-head performance or security benchmark.
- [1] Anthropic: Claude in Chrome general availability
- [2] Anthropic: Claude in Chrome permissions
- [3] Customaise: UserScripts
- [4] Customaise: AgentScripts and permissions
- [5] Customaise: Script Decisions
- [6] Customaise: plans and allowances
- [7] Customaise: privacy and data flows
- [8] Anthropic: shortcuts, scheduling and recordings
- [9] Anthropic: Claude in Chrome troubleshooting and usage
- [10] Customaise: on-device AI and your own models
Related comparisons
- Customaise vs Tampermonkey: Compare a userscript manager with Customaise’s integrated workflow for building, testing and reusing UserScripts and AgentScripts.
- Customaise vs Playwright: Playwright automates browsers from code. Customaise adds named tools and page controls to the Chrome you already use.
- Customaise vs Browser Use: Browser Use offers local and cloud paths, including Browser Harness. Customaise centres the workflow on an installed library of scripts and tools.