Gemini CLI

grip — Quizzes the developer on their own staged diff via a git hook and blocks the commit or push if they score below a pass mark.

deadcatfound — Runs a trading operations desk in demo mode, using a local CLI AI tool on the user's machine to process scheduled work.

vibe-kanban-alternative — Manages parallel git worktrees on a kanban board and gives coding agents persistent project memory across tasks.

holycode — Hosts a persistent cloud Linux machine with coding agent CLIs and dev tools preinstalled, accessible over SSH or browser terminal.

ai-discussion-room — Runs a multi-model deliberation where a chair LLM hears anonymized members' answers, using official CLI subscriptions instead of API keys.

ShadowWhispr — Turns speech into text as a dictation app that uses the user's existing Codex, Gemini or Claude CLI login instead of an API key.

Shumai — Provides video review, comments, asset processing, and team collaboration for creative work as an open-source alternative to Frame.io.

Quodeq — Scans source code locally with an LLM to flag security and quality issues mapped to CWE IDs and ISO 25010 dimensions.

Tempest — Runs multiple AI coding agents in isolated environments on a desktop, using a local code graph to help them find repository context with fewer tokens.

SPIRAL — Provides a desktop database client for querying and inspecting SQL Server, PostgreSQL, MySQL, SQLite, MongoDB, and Redis.

tah cartridges — Packs project files and documents into local binary index files so AI agents can retrieve relevant parts without sending everything to the model.

AI Debate Arena — Runs a multi-round debate between locally installed AI command-line agents and has a moderator AI summarize the discussion into a conclusion.

Vakh — Converts spoken dictation into text on the desktop, inspired by Wispr Flow.

Camael — Provides a terminal environment with coding agent support that runs without telemetry or cloud dependencies, for use on locked-down networks.

Laya — Aggregates notifications from work tools like Jira, Slack, Gmail and GitHub, links related items across them, classifies them with AI, and lets users stage replies and run agents on them.

dotpals — Displays plain-language summaries of what AI coding agents are doing on your desktop, including file changes, test failures, retries, and approval requests.

Shipabase — Sets up cookieless web analytics by having a coding agent read the codebase and propose a tracking plan for approval.

agents-context-router — Splits coding-agent instruction files into a small always-loaded kernel and task-specific topic pages, with a Python CLI to list, load, and check them.

Limitless — Keeps a Mac awake until a timer, a set time, or selected processes finish, with battery limits and hooks for AI coding agents.

CodeBurn — Tracks spending across AI coding tools by project, model and branch and flags unused setup that adds token costs.

Open Steps — Makes AI coding agents report progress in plain English, with done-or-not verdicts, required user actions, verification, and next steps.

shallnot — Checks that every requirement in a ticket has a passing test tagged with its ID, and sends coding agents back to work when one is missing.

Skiller — Manages AI coding-agent skills across multiple coding agents from one desktop view, with install, sync, edit, and remove actions.

Your Best Clip Is Buried In A… — Finds the best clips inside long videos so creators don't have to watch them in full.

CleeCode — Provides a terminal IDE that embeds coding agent CLIs like Claude, Codex, OpenCode, and Gemini beside the editor, with the agent seeing the cursor and applying edits to open buffers.

TokenNations — Ranks countries against each other by AI coding token usage, synced from local ccusage logs via a CLI.

AIByCity — Maps anonymized token totals from opt-in coding-agent usage by city and ranks cities.

sdlc-kit — Runs AI coding agents through staged gates, with human approval of specs and fresh-session agent review of built code.

Landline — Lets a coding agent session pause for a voice call with an AI that has read the session context and returns a decision memo.

I built a free tool that scans… — Scans a website for privacy, cookie-policy, and machine-checkable accessibility issues and reports fixes for the applicable laws.