JSPM

  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 83
  • Score
    100M100P100Q76919F
  • License MIT

aOa — semantic code search engine

Package Exports

    This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@mvpscale/aoa) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

    Readme

    aOa — Angle O(1)f Attack

    5 angles. 1 attack. 1 binary. Save your tokens. Save your time.


    CLI coding agents are inefficient. When Claude, Gemini, or any AI agent runs grep on your repo, it ignores .gitignore, scans everything — node_modules, build artifacts, vendored deps — and returns bloated, unorganized results. The agent then traverses that tree, reads every file, and burns through tokens you never see. A single grep can waste 100K+ tokens. They don't show you this because it doesn't look good.

    aOa exposes it. And fixes it.

    What it does

    aOa is a Go binary that sits in front of grep and egrep. When an AI agent calls grep, aOa intercepts it and returns precise, token-efficient results using an O(1) indexed lookup instead of brute-force file scanning.

    • O(1) token lookup — pre-indexed search, not line-by-line scanning
    • Respects .gitignore — never wastes tokens on files you already exclude
    • Drop-in replacement — agents don't know the difference, they just get better results
    • 95-99% token savings — on real codebases, measured, not estimated
    • Zero config — one binary, no Docker, no runtime deps

    Install

    npm install -g @mvpscale/aoa

    Setup

    # Index your project
    aoa init
    
    # Activate for AI tools — add to ~/.bashrc or ~/.zshrc:
    alias claude='PATH="$HOME/.aoa/shims:$PATH" claude'
    alias gemini='PATH="$HOME/.aoa/shims:$PATH" gemini'

    That's it. Your agent now uses aOa for every grep and egrep call.

    Commands

    Command What it does
    aoa init Index the current project
    aoa grep <pattern> Search with O(1) token lookup
    aoa egrep <pattern> Extended regex search
    aoa locate <path> Find files by path pattern
    aoa tree Project structure overview
    aoa health Check daemon and index status
    aoa daemon Run the background indexer

    Why "Angle of Attack"

    You're not seeing the true token economics of your AI tools. The waste is hidden at the lowest layer — the shell commands agents run on your behalf. aOa puts you back in control at that base layer, shielding you from token waste and giving you visibility into what's actually happening. That's the angle of attack.

    License

    Apache License 2.0. See LICENSE.