[GH-ISSUE #747] Add AI Input/Output System to VS Code Extensions (local-first routing, DeepSeek as cloud backend) #206

Open
opened 2026-10-26 09:55:06 +00:00 by bsakaguchi · 1 comment
Owner

Originally created by @1person280 on GitHub (Oct 1, 2026).
Original GitHub issue: https://github.com/deepseek-ai/awesome-deepseek-integration/issues/747

Project

Description

A local-first AI request router for VS Code. Simple prompts (for example: format this block, add a docstring, rename this variable) are answered on-device by a 4B INT8 model served through llama.cpp; only genuinely complex requests are forwarded to an OpenAI-compatible cloud endpoint.

DeepSeek can serve as that cloud endpoint by setting aiio.cloudBaseUrl to https://api.deepseek.com and aiio.cloudModel to deepseek-chat. In this setup paid DeepSeek calls are reserved for the requests that actually need them, and a built-in dashboard tallies the avoided cloud calls and the estimated cost savings.

Why it fits this list

It is a VS Code extension that drives the DeepSeek API (OpenAI-compatible) and would sit naturally under the VS Code Extensions section.

Disclosure

Author self-submission. Happy to open a PR with the README entry (and a docs/ai-input-output-system/ page) instead if you prefer that workflow.

Originally created by @1person280 on GitHub (Oct 1, 2026). Original GitHub issue: https://github.com/deepseek-ai/awesome-deepseek-integration/issues/747 ### Project - **Name:** AI Input/Output System - **Link:** https://github.com/1person280/ai-input-output-system - **Category:** VS Code Extensions - **License:** MIT ### Description A local-first AI request router for VS Code. Simple prompts (for example: format this block, add a docstring, rename this variable) are answered on-device by a 4B INT8 model served through `llama.cpp`; only genuinely complex requests are forwarded to an OpenAI-compatible cloud endpoint. DeepSeek can serve as that cloud endpoint by setting `aiio.cloudBaseUrl` to `https://api.deepseek.com` and `aiio.cloudModel` to `deepseek-chat`. In this setup paid DeepSeek calls are reserved for the requests that actually need them, and a built-in dashboard tallies the avoided cloud calls and the estimated cost savings. ### Why it fits this list It is a VS Code extension that drives the DeepSeek API (OpenAI-compatible) and would sit naturally under the **VS Code Extensions** section. ### Disclosure Author self-submission. Happy to open a PR with the README entry (and a `docs/ai-input-output-system/` page) instead if you prefer that workflow.
Author
Owner

@1person280 commented on GitHub (Oct 1, 2026):

Switched to the PR workflow: #749 adds this to the VS Code Extensions section. Please review the entry there, and feel free to close this issue once the PR is merged.

<!-- gh-comment-id:5923635604 --> @1person280 commented on GitHub (Oct 1, 2026): Switched to the PR workflow: #749 adds this to the VS Code Extensions section. Please review the entry there, and feel free to close this issue once the PR is merged.
Sign in to join this conversation.
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/awesome-deepseek-integration#206
No description provided.