Skip to content

Improve indexing and search for imported how-to docs - #3106

Merged
Hillary Mutisya (hillary-mutisya) merged 3 commits into
mainfrom
dev/hillarym/webImport
Sep 30, 2026
Merged

Hillary Mutisya (hillary-mutisya) merged 3 commits into
mainfrom
dev/hillarym/webImport

Conversation

@hillary-mutisya

Copy link
Copy Markdown
Collaborator

This pull request enhances the browser agent's web page indexing workflow to support reporting "how-to" extraction status and warnings, and improves the user experience in the browser extension by surfacing this information. The core changes add new interfaces and logic to check for "how-to" candidates after ingesting a page, propagate warnings and status up through the stack, and update the extension's context menu and UI feedback accordingly. Additional tests are included to verify these behaviors.

Enhancements to web page indexing and reporting:

  • Added support for reporting "how-to" extraction status and warnings when indexing a web page:
    • Extended the BrowserMemoryService.ingest method to optionally check for "how-to" candidates and return their status and count, including handling for unavailable services and warnings. [1] [2]
    • Updated the indexWebPageContent action and its service types to propagate howTo status, warnings, and errors to the caller. [1] [2] [3]

Browser extension improvements:

  • Added a "Save this page" option to the context menu, which triggers page indexing and displays status or warnings to the user via the badge and tooltip. [1] [2]
  • Modified the indexPageContent handler to return detailed indexing results, including "how-to" candidate info and warnings, and updated notification logic to reflect these results. [1] [2] [3] [4]

Testing and reliability:

  • Added comprehensive tests for the new "how-to" reporting logic, covering successful candidate detection, disabled detection, and error cases.
  • Improved error handling and propagation throughout the indexing workflow, ensuring errors and warnings are surfaced to the UI and RPC layers. [1] [2] [3]

Internal API and type changes:

  • Updated interfaces and type signatures throughout the agent and extension code to support the new result structure and options. [1] [2]

Dependency and construction changes:

  • Modified service constructors and adapters to support clients that implement the optional PersonalHowToService interface. [1] [2]

These changes collectively improve the visibility and reliability of the knowledge extraction process, especially around "how-to" content, for both developers and end users.

Add browser page saving with how-to candidate status and error feedback. Expand Claude memory search across conversations and saved-document corpora, and safely persist auto-generated procedure IDs.
Index latest procedure versions with KnowPro, including state tags, generation rebuilds, and corpus-scoped search. Expand reasoning memory search to query conversations, documents, and saved procedures in parallel with deduplicated evidence.

@datduyng Dominic Nguyen (datduyng) left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approve. Would be good if we can attach a screenshot of how it looks like

Merged via the queue into main with commit f739d46 Sep 30, 2026
27 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants