Инструменты разработчика
open-terminal
Open Terminal — a computer you can curl: a sandboxed terminal API for AI agents, from Open WebUI.
★ 3,3 тыс. звёздMIT
Смотреть руководство →destructive_command_guard — проект с открытым исходным кодом из категории «Инструменты разработчика» от Dicklesworthstone (Dicklesworthstone/destructive_command_guard на GitHub). У него 6 120 звёзд на GitHub и 253 форка, он распространяется по лицензии MIT и последний раз обновлялся на GitHub 8 октября 2026.
A guard that blocks dangerous git and shell commands before AI agents can execute them.
If an agent with shell access could wipe uncommitted work or damage your system, dcg is the seatbelt that sits in its command path.
Run the installer for your platform, using easy mode if you want sensible defaults.
Review the enabled packs, choose warn/log/ask/block behavior per rule, and add allowlist entries for your legitimate workflows.
Have your agent (or yourself) attempt a blocked command such as git reset --hard and confirm dcg intercepts it.
Intercepts dangerous git and shell commands before execution, explaining the reason and suggesting a safer alternative like git stash.
Organized packs cover git, filesystem, database, cloud and platform CLIs (including Azure DevOps), plus Windows-specific packs enabled by default on Windows.
Rules can block, warn, log silently or ask the operator for review where the hook protocol supports it, with severity ladders and presets such as strict_git.
Automatically detects which AI coding agent invoked it and applies per-agent disabled/extra packs and allowlists; Codex CLI is a first-class hook target alongside Claude Code.
An opt-in pack blocks value-emitting secret reads from tools like Infisical, 1Password, Doppler, Vault and AWS Secrets Manager.
dcg explain --format json shows exactly which rule and mode would apply to a command before you rely on it.
Clone the guard if you want the source; most users install the released binary in the next step.
git clone https://github.com/Dicklesworthstone/destructive_command_guard.git
cd destructive_command_guardRun the official installer in easy mode (a PowerShell installer is provided for Windows).
curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/destructive_command_guard/main/install.sh?$(date +%s)" | bash -s -- --easy-modeTry a destructive command through your agent hook and confirm dcg blocks it with an explanation.
git reset --hard HEAD~5
# dcg intercepts and blocks: git reset --hard destroys uncommitted changesThe Destructive Command Guard is maintained by Dicklesworthstone (Jeffrey Emanuel).
Yes — its LICENSE is the MIT License with an additional OpenAI/Anthropic rider; read the LICENSE file for the exact terms.
No. dcg is a local Rust binary that evaluates commands against rule packs; no network service or key is involved.
Permission prompts ask about every command or broad categories; dcg evaluates the actual command content against curated destructive-command packs and can block, warn or ask per rule.
Run the installer with --easy-mode, then test it by attempting git reset --hard through your agent.
Open Terminal — a computer you can curl: a sandboxed terminal API for AI agents, from Open WebUI.
Gmail for coding agents: identities, inboxes, threads, and file leases over FastMCP, Git, and SQLite.
One Rust CLI for every LLM: shell assistant, chat REPL, RAG, tools and agents across 20+ providers.
A lightweight, cross-platform code editor, built with Tauri (Rust and React) with Git support, AI agents, vim keybindings
A two-panel terminal UI for git diffs, with AI commits.
Stress-test your research paper with an AI reviewer before submission.
Другие проекты из категории «Инструменты разработчика» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если destructive_command_guard вам не подходит.
The open source coding agent
The model-definition framework behind modern AI — thousands of pretrained transformers for text, vision, and audio.
Anthropic's agentic coding tool that lives in your terminal and works across your whole codebase
Cross-platform desktop all-in-one manager for Claude Code, Codex, OpenCode and other coding agents
OpenAI's lightweight coding agent that runs locally in your terminal
Соседние проекты из категории «Инструменты разработчика» в каталоге, ближайшие по числу звёзд к destructive_command_guard — продолжайте изучать полку по обе стороны от него.
Fast, accurate semantic code search for agents — about 99% fewer tokens than grep plus read.
The CLI and skills that turn any coding assistant into an expert at creating, evaluating, and deploying AI agents on Google Cloud.
An AI-powered security review GitHub Action using Claude to analyze code changes for security vulnerabilities.
It's like v0, but in your Cursor / Claude Code / Windsurf: search 10,000+ React/Tailwind components, generate new UI with AI, and publish your own — right fr…
Self-hosted control plane for AI agents: dispatch tasks, review runs, track spend, and operate OpenClaw, Claude Code, Codex, and other runtimes.
A context graph for the AI-native SDLC — so agents and humans share one map of the codebase.
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.