Smooth Browser
ReviewAudited by ClawScan on May 10, 2026.
Overview
Smooth Browser is a coherent browser-automation skill, but it should be reviewed because it can reuse logged-in web sessions through an external agent with broad authority and limited required scoping.
Install only if you are comfortable using Smooth as an external browser automation service. Use separate profiles per website, restrict sessions with --allowed-urls, avoid sending unnecessary personal data or files, and require explicit confirmation before the agent submits forms, posts content, or changes account data.
Findings (5)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If used incautiously, the agent could submit forms, post content, or change account data while logged in.
The documented workflow shows a natural-language browser agent reusing an authenticated profile to perform an account-mutating action. The shown instructions do not require a confirmation step, URL allowlist, or read-only mode before such actions.
smooth start-session --profile-id "github-account" smooth run -- <session-id> "Create a new issue in my repo 'my-project'"
Use explicit user approval before submissions or account changes, prefer --allowed-urls for each task, and use --profile-read-only when mutation is not needed.
A saved profile may let future browser tasks act as the user on websites without re-authentication.
Persistent browser profiles can carry authenticated account authority across sessions. This is useful for the skill's purpose, but it is high-impact and the artifacts do not clearly bound retention, revocation, or which services should be reused.
Profiles are useful to persist cookies, login sessions, and browser state between sessions.
Create separate profiles per site, delete profiles when no longer needed, and require user confirmation before reusing an authenticated profile.
The agent may remember and reuse logged-in browser profiles across tasks, which can expose or misuse account access if the memory is wrong or over-trusted.
The skill explicitly asks the agent to store persistent mappings between profile IDs and authenticated services. That memory could be stale, poisoned, or reused later without enough user context.
Save to memory which profiles authenticate to which services so you can reuse them efficiently in the future.
Store only minimal profile labels, avoid saving sensitive account details, and ask the user before reusing any authenticated profile.
Sensitive task details, browsing actions, and metadata may be processed outside the local assistant.
The host agent is delegating web tasks to another intelligent agent. This is central to the skill, but users should understand that task instructions and browsing context may be handled by the Smooth agent/service.
IMPORTANT: Smooth is powered by an intelligent agent, DO NOT over-controll it
Do not pass secrets or private data unless necessary, and keep tasks narrowly scoped to the intended site and action.
Installing the package gives local execution capability to code outside this skill artifact.
The skill relies on an external, unpinned package installed by the user, while the reviewed artifact set contains no code files. This is normal for a CLI integration, but the runtime implementation is not reviewed here.
pip install smooth-py
Install from the official Smooth source, verify the package identity/version, and keep it updated.