[Feature]: Add a persistent Find in Files workflow

[kyrylofranko]

Before submitting

• I searched existing issues and Ideas discussions and did not find a direct duplicate.
• I am describing a concrete problem and use case.

Area

apps/web - project content search on web and desktop.

Problem or use case

T3 Code already has useful project content search with case, whole-word, and regex controls. The current search is a transient modal. Opening a match closes the result list, and each result shows only its matching line.

This is one of the few workflows that still sends me back to WebStorm. During code exploration, I need to refine a search, compare matches across files, inspect the surrounding code, and return to the same result set without rebuilding my context.

JetBrains Find in Files is a good workflow reference: https://www.jetbrains.com/help/webstorm/finding-and-replacing-text-in-project.html

The most useful parts are file masks, search scope, recent search patterns, a full code preview, and persistent result tabs. I do not think T3 needs to copy the JetBrains UI. The workflow is what matters.

Proposed direction

Make project content search usable as a persistent code exploration surface:

• Keep the result list available while I select and inspect matches.
• Show the surrounding file in a preview at the selected match.
• Add include and exclude file masks, such as *.ts,*.tsx,!*.test.ts.
• Allow project or directory scope.
• Remember recent query, mask, and option combinations. Named saved searches can come later.
• Optionally open a result set in a persistent search tab when I need to compare multiple searches.

Smallest useful scope

I would start with persistent results plus a code preview, while keeping the current query and options available. File masks look like the next useful slice. Search scopes, named searches, and multiple result tabs can stay separate.

Why this matters

T3 already covers the agent, terminal, files, diffs, and browser preview. Better content search would close one of the main gaps that still requires a separate IDE during an agent workflow.

Alternatives considered

Keep using WebStorm for repository-wide search. It works well, but it breaks the T3 workflow and requires another application for routine code exploration.

Screenshots

I did not include screenshots because my real examples contain private production code. The public JetBrains documentation above demonstrates the reference workflow without exposing project data.

Questions

1. Does this direction fit the intended scope of T3 Code?
2. Would a persistent right-panel surface or a dedicated search tab fit the architecture better?
3. Which first slice would you accept from an external contributor?

Contribution

• I would be open to helping implement this.