Skip to content
| Marketplace
Sign in
Visual Studio Code>AI>Code Context by dotnetpowerNew to Visual Studio Code? Get it now.
Code Context by dotnetpower

Code Context by dotnetpower

Preview

dotnetpower

|
2 installs
| (0) | Free
Fast local code search, source reads, and context for AI agents. Native VS Code tools backed by one shared, adaptive cache.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Code Context — local context, less friction

Code Context by dotnetpower

Native VS Code source context for your coding agent on Linux, native Windows (cctx.exe), WSL, and SSH. Find files, search source, and read block-clamped nearby code with seven native VS Code Language Model Tools (#cctxNativeContext, #cctxNativeGrep, #cctxNativeRead, #cctxNativeFiles) backed by one shared Rust cache, <8 KiB inline response guards, rich document anchors (#toc, #slide-N-M, #heading), and live LLM model/token/AIC/USD cost savings telemetry. Publisher: dotnetpower. Previous VSIX users: migration steps.

Resources

Repository · Setup · Tool reference · Benchmark report · Issues · License

VS Code's native right-hand Resources panel uses Marketplace metadata. For VSIX-only installations, use the links above instead.

Performance & cctx usage example

$ cctx usage --hours 2        # or: cctx usage --hours 24 --json
⚡ CODE CONTEXT (cctx) — MODEL, TOKEN & COST SAVINGS DASHBOARD (Window: 2.0h)
💰 COST SAVED (@$0.001/AIC) : +$2.42 USD (-28.4% of baseline bill · ~2421.6 AIC saved)
💎 API LIST VALUE (@$0.01)  : +$24.22 USD | 🔥 TOKENS SAVED: ~86.02M tok (244 Turns + 35x Cache + BPE)
🎯 WARM CACHE HIT RATE      :  98.6% (506/513) | 🚀 ELIGIBLE SOURCE ADOPTION: 98.5% (513/521)

On AKS Store Demo, the fixed retrieval mix averaged 14.92× faster with a warm CCTX cache (95% CI 14.17–15.80×, up from 6.83× in v0.14): 4.246 → 0.284 ms per operation, with a 28.675 ms cold sync median (down from 98.813 ms in v0.14). Replacing separate rg + view calls with one block-clamped code_context call cuts lookup payload tokens by 73.2–91.3%, rich document (#toc / #slide-N) reads by 86.3–98.9%, and reduces end-to-end coding agent wall time by 24.7%.

Task CCTX off CCTX on, warm Off / on
Filename discovery 4.724 ms 0.304 ms 15.54×
Literal search 5.574 ms 0.273 ms 20.41×
Search + context 5.680 ms 0.319 ms 17.80×
Known read + hash 0.085 ms 0.234 ms 0.36× — slower
Fixed mixed suite 4.246 ms 0.284 ms 14.92×
Token & agent acceleration suite Baseline (rg + view) With CCTX (code_context / code_read) Savings / speedup
Rust 6-symbol lookup tokens 15,028 tok (100L) / 25,133 tok (200L) 4,031 tok (6 calls vs 20, 10.3× faster) -73.2% to -84.0% tokens
Python 6-symbol lookup tokens 13,997 tok (100L) / 24,609 tok (200L) 2,141 tok (13.16× faster) -84.7% to -91.3% tokens
Rich documents (.pptx/.docx/.md #toc & #slide-N) Full-document read Virtual #toc + #slide-N-M (5.4× faster DEFLATE) -86.3% to -98.9% tokens
Copilot A/B agent turns & cost 7.00 calls / 5.00 turns / 20.31 s 2.67 calls / 3.67 turns / 15.30 s -50.0% tok · -25.4% AIU · -24.7% time
Live status bar & cctx usage telemetry Unmonitored Per-model tokens, 98.6% warm hit & AIC/$USD savings -10% to -39% session bill
Auto-loaded instructions & cctx prompt 9,127 B / 2,048 tok prose BPE-aligned formula spec (756 tok, EN/KO --compress) -63.1% instruction tokens

Warm retrieval latency, speedup, Rust/Python token savings, and agent acceleration

Eight daemon sessions and 8,960 correctness-checked pairs; OS cache warm in both arms. Known small reads were slower through CCTX. Cold misses fall back instead of waiting. Full method, cold-start chart, raw data, and reproduction

Install

Preview · Local backend required. Supports native Windows (cctx.exe, no WSL needed), Linux, WSL, or Linux SSH with stable Rust and ripgrep. VS Code 1.140+ is required; agent instructions (applyTo: "**") attach automatically.

1. Install the backend (Linux/WSL/SSH or native Windows):

git clone https://github.com/dotnetpower/code-context.git && cd code-context
bash scripts/install.sh && ~/.local/bin/cctx ping
# Native Windows (PowerShell): cargo build --locked --release; copy target\release\cctx.exe $HOME\.local\bin\cctx.exe

2. Download the VSIX. In VS Code choose Extensions → … → Install from VSIX….

3. Run Code Context: Getting Started in a trusted workspace (hover the CCTX N status bar item or run cctx usage --hours 24 to view live LLM model, token, cache hit, and AIC/$ cost savings). On Windows local folders (C:\...), cctx.exe daemon starts automatically in the background:

Use #cctxNativeFiles to find *Service*, then #cctxNativeContext for nearby source.

No proposed API flags are needed. Marketplace publication is separate from the downloadable VSIX. Windows, WSL, and troubleshooting details are in the setup guide. The cache is optional: CACHE_MISS is not an empty search; continue with permitted host tools.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
  • Your Privacy Choices
  • Consumer Health Privacy
© 2026 Microsoft