Browser Commands are Chrome-aware buttons on the desktop Power Strip. They are not a separate browser strip and not an in-page menu.
Visibility and Connection
Power Strip shows browser commands when Chrome is the foreground application. Foreground detection and extension connectivity are separate:
- Chrome in the foreground makes the buttons visible.
- An active extension connection is required to execute them.
- A visible button can therefore return "Chrome not connected."
The buttons carry a browser badge to distinguish them from desktop-only commands.
Command Catalog
Save Chat
Asks the active tab for its conversation and sends the result to the desktop import pipeline. A Save Chat slot has its own settings for how the result is saved and which metadata fields are included.
Tag Page
Runs the save-with-tag flow for the active page or conversation.
Capture Page
Captures the active page. Its per-slot settings include:
- Capture mode: clean scrape, snapshot, or metadata only
- Save mode: Markdown, Markdown plus JSON, or metadata only
- Optional metadata fields
On supported chat sites, capture can use the platform's conversation structure rather than treating the page as a generic article.
Capture Selection
Captures the current browser selection and imports it into the active ThinkForge project.
Capture Page and Capture Selection can trigger an after agent hook once the imported document is available.
Previous Q&A and Next Q&A
Moves through question-and-answer pairs, using platform-specific navigation where available.
When invoked from Power Strip these still require the extension connection, because the desktop must send the command to the active tab.
Personas
Opens the extension's Personas surface.
Template
Opens the extension's Forge Doc template surface.
Dockit
Not functional in this beta. The button is present, but the desktop-triggered action reports "Dockit module not yet available."
Browser Agent
Opens a Power Strip sidecar for selecting and configuring an agent. At run time, the desktop requests content from the current browser page and runs the bound agent without further input.
Known gap: the sidecar displays reference-document and output-folder configuration, but those stored values are not currently loaded into execution. Do not rely on them as effective inputs or output routing yet.
Run Automation
Requests current-page context from Chrome and runs it through the bound automation.
Per-Slot Settings
Save Chat and Capture Page expose a browser settings sidecar, stored against that individual command slot. Other browser commands do not use that panel.
Data Flow
There are two directions:
- Desktop to extension — A Power Strip click sends an action to the extension, which routes it to the active tab or an extension window.
- Extension to desktop — Captured or saved material returns to the desktop and enters ThinkForge's normal project import pipeline.
Everything travels over localhost:19222.
Capture Completion, Hooks, and Preview
Browser capture completion is reported back to the desktop after import. That completion can:
- Supply the saved document to an after hook
- Request Preview on Save
- Update feedback
These behaviors depend on the complete round trip, not merely on the initial Power Strip click.
Requirements
Browser Commands require:
- ThinkForge desktop running
- The Think Forge Chrome extension installed
- A connected extension
- An active browser tab that supports the requested operation
Content extraction and navigation quality depend on the active site and its current page structure.