JSPM

  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 31
  • Score
    100M100P100Q61694F
  • License MIT

AI coding assistant for your terminal

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 (glitool) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

    Readme

    glitool

    AI coding assistant for your terminal. Multi-agent pipeline, smart routing, and a live process trace — all without leaving your terminal.

    Install

    npm install -g glitool

    Setup

    mkdir -p ~/.glitool
echo "OPENAI_API_KEY=sk-..." > ~/.glitool/.env

    Then start:

    glitool

    Slash Commands

    Command Description
    /plan Create a structured plan for a complex task
    /coder Run the full multi-agent coding pipeline
    /debug Diagnose errors and broken behavior
    /refactor Restructure code without changing behavior
    /review Audit code for bugs, security, and quality
    /git Commit, push, diff, branch — full git operations
    /explain Explain a concept or file (no file edits)
    /quick Fast chat, cheapest model, no pipeline
    /model Show or switch the active model
    /memory View project memory and session summary
    /tools List available tools
    /clear Clear session (keeps memory)
    /reset Clear session and wipe memory
    /exit Save summary and quit

    Smart Routing

    You don't need slash commands. Glitool reads your message and picks the right agent automatically:

    why is my server crashing?       → DEBUGGER
review src/auth.ts               → REVIEWER
refactor the parser module       → REFACTORER
commit my changes                → GIT AGENT
how does useEffect work?         → EXPLAINER
add a rate limiter               → CODER (full pipeline)

    Multi-Agent Pipeline

    For coding tasks, four agents run in sequence:

    PLANNER → CODER → VALIDATOR → JUDGE
    • PLANNER reads your request and produces a numbered step-by-step plan
    • CODER executes the plan using file and shell tools
    • VALIDATOR runs TypeScript and ESLint checks on the result
    • JUDGE reviews the output and decides if it meets the requirement

    Each stage is shown live in the terminal as it runs — with reasoning text and every tool call displayed in order.

    Tools

    Agents have access to:

    Tool What it does
    readFile Read any file in the project
    listFiles List files matching a glob pattern
    searchCode Search source files for a string or pattern
    writeFile Create a new file
    editFile Edit an existing file
    bash Run shell commands (risk-gated)
    webFetch Fetch a URL and read its content

    Dangerous shell commands (rm -rf /, sudo, curl \| sh) are blocked. Sensitive commands (git push, npm publish) require your confirmation.

    Memory

    Glitool remembers context across sessions:

    • Session memory — last 40 messages saved per project, auto-summarized
    • Project memory — tech stack, architecture decisions, and TODOs extracted from your conversations, stored in .glitool/memory.json

    Configuration

    Config file: ~/.glitool/config.json

    {
  "name": "Developer",
  "preferredLanguage": "TypeScript",
  "codingStyle": "spaces",
  "preferredModel": "gpt-4o-mini"
}

    Requirements