ClaudeMap

Geiger

A read-only one-command inventory of every AI agent, harness, MCP server, plugin, and AI extension on a machine, with plain-language exposure labels.

SDKs & Toolingsecurityauditinventoryclinpm
Visit projectAtomburstofficial/geiger

Geiger answers the question most people cannot: what is actually running on this machine, and what can it reach? One read-only command scans configs and dotfiles, groups findings by ecosystem, and labels what each item can execute, read, and hold — including credentials sitting in config files and wrapper/policy-agent relationships. Nothing is written unless you ask for --json output.

What it does

  • npx geiger-scan — no install, no account, no telemetry; reads configs and directories, writes nothing by default
  • Findings grouped by ecosystem with capability labels: EXECUTES, HOLDS-SECRETS, BROAD-FILESYSTEM, NETWORK
  • Flags credentials under credential-named keys in files like ~/.claude.json (displayed redacted) and notes policy-agent wrappers
  • Zero dependencies, Node ≥ 18, MIT; --json only when you explicitly ask for a file

When to use it

For anyone whose machine has quietly accumulated agent harnesses, MCP servers, and plugins across months of experimentation: audit what each can reach before installing the next one, or review a machine before handing it access to production credentials.

Getting started

npx geiger-scan

# persist the report (the only thing it ever writes):
npx geiger-scan --json geiger-report.json

Things to know

Geiger is an inventory, not a defense: it tells you what is running and what it can touch, but removing or sandboxing anything is on you. Labels are heuristic reads of config files — a wrapped or proxied server can still surprise you, so treat results as a starting point for review.

Repository

Atomburstofficial/geiger

Charted

Related on the map

  • Harness Bridge

    SDK / Tool

    A local launcher that points Claude Code and seven other coding harnesses at any OpenAI-, Anthropic- or Responses-compatible endpoint — without editing a single config file.

    • launcher
    • local-models
    • cli
    • menu-bar
    View

  • Claude for Siri

    SDK / Tool

    Answers macOS Spotlight and Siri prompts with your signed-in Claude Code account — an experimental macOS 27 model-delegation provider.

    • macos
    • siri
    • spotlight
    • model-delegation
    View

  • AI Coding Assistant Data Extractor

    SDK / Tool

    Extract your own local chat history from ten AI coding assistants into one normalized JSONL — for fine-tuning, analytics, or backing up years of conversations. Standard library only.

    • cli
    • data-export
    • jsonl
    • backup
    View