|
Termaxa — a gate for the shell commands AI coding agents run
The agent's prompt shows you the command; Termaxa shows you the consequence. It previews the blast radius, backs up first, blocks the dangerous ones, and kee...
|
84 % |
|
Rubric — Let Claude Code run unattended. We block what’s dangerous.
Stop clicking approve on every Claude Code permission prompt. Rubric lets the agent run autonomously and blocks only the truly dangerous stuff — rm -rf /, fo...
|
83 % |
|
CC Safety Net - Block Destructive AI Coding Agent Commands
A coding agent CLI safety layer that blocks destructive git and filesystem commands and secret file access before execution.
|
83 % |
|
Kintsugi — Safety layer for AI coding agents | Stop rm -rf, git push --force, DROP TABLE before they run
Kintsugi is a local-first safety layer and command firewall for AI coding agents — Claude Code, Cursor, Codex, Gemini, Copilot. It catches dangerous shell co...
|
82 % |
|
CodeCave — Self-hosted AI coding agent orchestration
Run a team of AI coding agents on your own machine. Isolated git branches, merges you approve, guard rails — your code never leaves your box.
|
82 % |
|
oops — Undo for your terminal | Ctrl+Z for the command line
oops is an open-source CLI that backs up files before destructive terminal commands like rm, mv, sed -i, and git reset --hard. Type oops to restore them inst...
|
82 % |
|
Offsend — One file defines your AI context boundary
Open-source, local-first boundary layer for AI coding agents. One file in git. After install, secrets become tokens — the agent keeps working. Ignore, gates,...
|
82 % |
|
clonecode — parallel Claude Code instances on isolated clones
Spin up isolated git clones and launch Claude Code in each one. Run multiple agents in parallel without them stepping on each other's working trees.
|
82 % |
|
wardayacode — AI coding agent for your terminal
An open-source, multi-provider AI coding agent that lives in your terminal. Claude, GPT-4, and Gemini.
|
82 % |
|
softwareaifactory.sh — enforcement you can execute
A software factory template for AI coding agents. Deterministic gates, role separation, and a verification contract — every control an executable script, eve...
|
82 % |
|
Git Shipit
Shipit is an open source command line interface for managing merge requests, changelogs, tags, and releases using a plan and apply interface. Built with codi...
|
81 % |
|
Dangerous Sandbox — agents, off your machine
An isolated environment where AI agents can edit files, run code, and experiment — without touching your machine. Remote, kernel-isolated containers with str...
|
81 % |
|
codekeel, drift governance for AI coding agents
Drift governance for agentic coding. Record the decisions your AI agent makes, and have them enforced automatically on every future turn.
|
81 % |
|
dotboost — Claude Code configuration for Dotkernel projects
dotboost is a drop-in .claude/ directory that teaches Claude Code the conventions of any Dotkernel application — API, Admin, Frontend, Light or Queue. Permis...
|
80 % |
|
WormBytes
|
80 % |
|
Brakelog: flight recorder and kill switch for coding agents
Open-source Claude Code hook that logs every tool call to a hash-chained log it protects from the agent, blocks credential reads and risky commands, and adds...
|
80 % |
|
codeout - your AI coding agents, self-hosted, anywhere
codeout is a self-hosted GUI for your AI coding agents. Run Claude, Codex, or Gemini on your own machine and drive it from your phone or browser, as a chat o...
|
80 % |
|
devme | the executable README
One command to go from git clone to a running dev environment. devme reads your project config, installs dependencies, starts services, and drops you into a...
|
80 % |
|
AgentFactory — an agent multiplexer built on tmux
An agent multiplexer built on tmux: run Claude Code, Codex, Grok, OpenCode — or any command — as managed sessions with live status, agent-to-agent coordinati...
|
80 % |
|
Archcore — Spec-Driven Development & Context Engineering
Spec-driven development and context engineering for AI coding agents. Keep specs, architecture, decisions, rules, and plans versioned with code.
|
80 % |
Detected social media networks