Read the project before answering
Use the active file and relevant project context to ground guidance in the code the engineer is actually changing.
PROJECT-AWARE RETRIEVAL · ACTIVE-FILE CONTEXT
Embedded Agents overviewVS Code implementation
Project-aware assistance for understanding code, drafting changes, and reviewing the result—delivered through the development surface your team already uses.
EXPLORER
▾ src
▾ services
parser.ts
▾ utils
stream.ts
tests
REQUEST
Refactor the parser to use the streaming pattern already established in this project.
PROJECT CONTEXT FOUND
Matching helper identified. A reviewable refactor and companion test update are ready.
@@ parser input handling
- const data = await fetchAll(input);
+ const data = await stream.read(input);
+ return parseChunks(data);
In the editor
No separate AI workspace
Project-aware
Context follows the code
Review first
Diff before apply
Team-shaped
Curated prompts and tools
Customer-hosted
Your chosen environment
An editorial IDE workflow
Use the active file and relevant project context to ground guidance in the code the engineer is actually changing.
PROJECT-AWARE RETRIEVAL · ACTIVE-FILE CONTEXT
Draft edits and supporting tests with a visible diff so the engineer stays in control before anything is applied.
REVIEWABLE EDITS · TEST GENERATION
Explain changes, summarize diffs, and help prepare the communication that surrounds a code review.
CHANGE EXPLANATION · DELIVERY SUMMARIES
Demonstration
EXPLORER
▾ src
▾ services
parser.ts
▾ utils
stream.ts
tests
REQUEST
Refactor the parser to use the streaming pattern already established in this project.
PROJECT CONTEXT FOUND
Matching helper identified. A reviewable refactor and companion test update are ready.
@@ parser input handling
- const data = await fetchAll(input);
+ const data = await stream.read(input);
+ return parseChunks(data);
The assistant can bring project context, a proposed change, and the review surface together without replacing engineering judgment.
Organizational fit
ENGINEERS
Ask about the project, explore unfamiliar code, and review proposed changes without moving into a separate AI workspace.
PLATFORM TEAMS
Shape approved prompts, knowledge, and tools around the standards your engineering organization already follows.
SECURITY & IT
Deploy within the chosen environment and govern access, model choice, and usage through the same platform foundation.
Deliver the extension through the channel that fits your organization and keep the experience aligned with enterprise policy.
Run the supporting platform in your chosen cloud, on-prem, or through a disconnected deployment pattern.
Connect the project and organizational knowledge the team is authorized to use without publishing an internal systems inventory.
Bring a real repository
KEEP EXPLORING
VS Code is one Embedded Agents implementation.