D
DevcoreAI

Getting Started

Adding Context

Bring files, errors, terminal output, and web content into any message using @mentions or drag & drop.

@Mention Reference

All supported @mention types

@/path/to/file

Include the full content of a specific file.

@/src/api/users.ts
@/path/to/folder/

Include all files in a folder (recursively).

@/src/components/auth/
@problems

Attach all diagnostics from the Problems panel.

@problems Fix these TypeScript errors
@terminal

Attach the current terminal output.

@terminal The build is failing. Why?
@git-changes

Attach the current unstaged / staged diff.

@git-changes Review my changes
@<commit-hash>

Attach the diff of a specific Git commit.

@a1b2c3d What did this commit change?
@https://example.com

Fetch and attach the content of a URL.

@https://react.dev/learn

Combining Mentions

You can mix multiple @mentions in a single message. DevcoreAI collects all the referenced content before sending to the model.

Example — debugging with full context

I'm getting these errors: @problems
Here's my component: @/src/components/Form.jsx
And the API route: @/src/api/users.js
The error started after: @a1b2c3d
And terminal says: @terminal

Drag & Drop

Drag any file directly into the DevcoreAI chat input. Supported types:

Source files

.ts, .tsx, .py, .js, .go, .rs — any text file

Images

.png, .jpg, .webp, .gif — attach visual context

Text/JSON

.txt, .json, .yaml, .md — configs, docs, logs

PDFs

PDFs are read and text content is extracted

Editor Context Menu

Right-click any selected code in the editor to send it directly to DevcoreAI:

  • Add to DevcoreAIAdds the selection to your next message as context
  • Explain with DevcoreAIOpens a new task asking the agent to explain the selected code
  • Fix with DevcoreAIOpens a new task asking the agent to fix the selected code
  • Improve with DevcoreAIOpens a new task asking the agent to improve the selected code

Terminal Context

You can also right-click selected terminal output and choose Add to DevcoreAI to paste it directly into your message. Combined with @terminal, this gives the agent full visibility into runtime errors and build output.