Profile
Back to NewsBack
GitHub Trending 4 min
Reader Mode
Domush/antigravity-plugin-context-mode: Reduce context size by 90%+ *and* make your agent smarter. Antigravity Context Mode is a local-first plugin built specifically for the Google Antigravity App. It prevents terminal commands, build logs, and mult

Domush/antigravity-plugin-context-mode: Reduce context size by 90%+ *and* make your agent smarter. Antigravity Context Mode is a local-first plugin built specifically for the Google Antigravity App. It prevents terminal commands, build logs, and mult

Antigravity Context Mode

Stop context bloat. Save 90%+ of your prompt tokens. Keep your agent fast, focused, and sharp.

npm version</a> Live Simulator</a> License: MIT</a> ![Zero Runtime Dependencies]() ![Antigravity App Compatible]()

Antigravity Context Mode is a free, local-first plugin built specifically for the Google Antigravity App. It prevents terminal commands, build logs, and multi-megabyte source files from flooding your conversation history — saving thousands of tokens per turn and stopping "agent amnesia" in its tracks.

🌐 Interactive Live Simulator & Documentation: domush.github.io/antigravity-plugin-context-mode
image

Quick Install (10 Seconds)

Install directly into Google Antigravity using npx:

# Global install (available across all Antigravity chats)
npx antigravity-plugin-context-mode install

Or per-project install (scoped to current workspace only)

npx antigravity-plugin-context-mode install --workspace

Restart or open the Google Antigravity App, and Context Mode is active immediately.


The Problem: Context Window Bloat

When an AI coding agent executes tools, standard workflows dump everything directly into your conversation context:

  • Running Get-Content Logs/Editor.log dumps 65 KB (~16,000 tokens) into your prompt.
  • Running test suites, compiler outputs, or git diffs dumps 100+ KB of repetitive text.
  • Reading large files or documentation dumps thousands of lines of irrelevant boilerplate.

Why this hurts:

  1. Agent Amnesia: Important system instructions, user constraints, and earlier architectural decisions get drowned out.
  2. Slower Responses: Generation times climb as the context window fills up with raw text.
  3. Burned Quota: You hit model context ceilings and rate limits after only a few turns.

How It Works

┌────────────────────────────────────────────────────────┐
│ 1. Agent attempts to read a large file or command      │
│    (e.g. 5,000-line build log or 40 KB source file)    │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
┌────────────────────────────────────────────────────────┐
│ 2. Pre-Tool Lifecycle Hook Intercepts Unbounded Dumps  │
│    • Files under 8 KB pass through normally            │
│    • Un-paged dumps over 8 KB are safely intercepted   │
│    • Images, docs, and chat attachments are exempted   │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
┌────────────────────────────────────────────────────────┐
│ 3. Sandboxed Execution ("Thinking in Code")            │
│    • Agent filters data locally using Node/Python      │
│    • Indexes text into local SQLite FTS5 BM25 database │
│    • Returns only the exact answer or error to chat!   │
└────────────────────────────────────────────────────────┘

Proven Results

Tested live on real-world build logs and source files in the Antigravity App:

| Operation | Without Context Mode | With Context Mode | Savings | | :--- | :---: | :---: | :---: | | Parsing 665-line Unity Editor.log | 65,242 B (~16,300 tokens) | 550 B (~137 tokens) | 99.2% saved | | Full File Read on 40 KB component | 40,960 B (~10,240 tokens) | Intercepted & paged to 50 lines (~800 tokens) | 92.1% saved | | Searching Compiler Warnings | Entire log dumped into chat | Top 2 ranked FTS5 snippets (~350 tokens) | 97.8% saved |


Key Features

  • 🛡️ Automated Interception: Blocks unbounded file dumps and commands before they bloat your context window.
  • ⚡ Thinking in Code: Filters multi-thousand-line outputs in 50 ms via local sandboxed scripts.
  • 🔍 Local SQLite FTS5 Search: Indexes files and logs locally on your machine with BM25 keyword matching and trigram search.
  • 📊 Built-In Sidecar Dashboard: Real-time stats pane inside Antigravity showing tokens kept out of context.
  • 🔒 100% Local & Private: All indexed data stays on your machine in ~/.context-mode/context.db. Zero network calls.
  • 📦 Zero Dependencies: Pure native Node.js (node:sqlite). No native builds, no node-gyp.

Available MCP Tools

Once installed, Context Mode provides 7 specialized tools for your agent:

| Tool | What It Does | | :--- | :--- | | ctx_execute | Runs sandboxed Node.js, Python, or PowerShell scripts to inspect files and extract answers without context dumps. | | ctx_search | Searches files, directories, or URLs with Just-In-Time auto-indexing using local SQLite FTS5 BM25 + Trigram RRF. | | ctx_batch | Runs multiple terminal commands in batch and indexes the combined output. | | ctx_stats | Displays a formatted context savings report and token metrics. | | ctx_resume | Returns recent indexed sources and state pointers after context compactions. | | ctx_doctor | Checks SQLite database health, WAL mode, and platform runtimes. | | ctx_purge | Clears stored sources and events from SQLite. |


Managing the Plugin

Enable / Disable Anytime

Open the Google Antigravity App → Customizations → Plugins, and toggle Context Mode on or off with one click.

Adjust Cutoffs

You can adjust the file read cutoff directly in the sidecar dashboard dropdown (2 KB, 4 KB, 8 KB, 16 KB, 32 KB, 64 KB, or none).

Uninstall

To remove the plugin from your system:
# Remove global plugin
npx antigravity-plugin-context-mode uninstall

Remove per-project plugin

npx antigravity-plugin-context-mode uninstall --workspace

License

MIT © Phillip Webber

Chat with me