simion/termic

Open-source Conductor.build alternative. Runs the real claude/gemini/codex CLIs in real terminals, no SDK middleman

What it solves

Termic is a desktop application designed to manage multiple AI coding agents (like Claude Code, Codex, and others) simultaneously. It solves the problem of running agents in isolation, avoiding the need to manually manage multiple git worktrees or deal with vendor-hosted backends and separate SDK billing. It allows developers to run several agents in parallel to compare their outputs or brainstorm architectural decisions without interfering with their main codebase.

How it works

Termic spawns the actual CLI binaries of AI agents rather than using SDKs, meaning it leverages the user's existing Pro/Max subscriptions. Each agent runs in its own real PTY (pseudo-terminal) and is isolated in a dedicated git worktree. For security, it offers two isolation methods: a macOS-specific Seatbelt sandbox (combining filesystem and network restrictions via an in-process proxy) and a cross-platform Docker mode for stronger filesystem boundaries.

Who it’s for

It is built for developers who use AI coding agents and want a high-degree of control over their environment, security, and cost, specifically those who prefer open-source tools and want to run agents in parallel across multiple repositories.

Highlights

  • Parallel Worktrees: Run multiple agents on the same branch in isolated git worktrees.
  • Broadcast & Brainstorm: Send a single prompt to all active agents concurrently to compare results.
  • Broadcast & Brainstorm: Send a single prompt to all active agents concurrently to compare results.
  • Broadcast & Brainstorm: Send a single prompt prompt to all agents concurrently to compare results.
  • Multi-repo Tasks: Group several repositories under one task with shared configuration.
  • Integrated IDE Features: Includes a fuzzy file finder, CodeMirror 6 editor with diffs, and language server support for code navigation.
  • Integrated Git UI: A Fork-inspired UI for staging, committing, and viewing a commit graph.
  • Security Sandboxing: Per-task isolation using macOS Seatbelt or Docker containers.
  • Spotlight: Real-time mirroring of worktree changes into the repo root without committing to the branch.
  • Custom Agent Registry: Support for built-in agents like Claude and Grok, or custom PTY-based agents like aider or ollama.

Related

  • Project
    praetorian-inc/augustus

    A Go-based LLM vulnerability scanner that tests models against 210+ adversarial attacks, including prompt injection and jailbreaks, to identify security weaknesses.

  • Project
    open-thoughts/OpenThoughts-Agent

    OpenThoughts‑Agent (OT‑Agent) is a research‑focused toolkit for generating training data, fine‑tuning, and evaluating small agentic language models. It bundles a datagen pipeline (Ray + vLLM), SFT support via LLaMA‑Factory, and a Harbor‑based evaluation framework, with launchers that work on SLURM clusters, cloud VMs (SkyPilot), or locally. The codebase is aimed at researchers building and benchmarking agentic LLMs.

  • Project
    About-Intelligence/soku-cli

    Soku CLI is a Node.js command‑line tool that gives AI agents secure, typed access to a full marketing/analytics stack (ads platforms, GA4, PostHog, SEO hosting, automations, etc.). It authenticates via Soku, returns stable JSON for non‑interactive use, and gates any write operations behind a human review step. Install with npm, log in, select a workspace, install the needed "skills", and then run commands like `soku ads query‑single‑dimension …` or `soku automation create …` to let agents retrieve data, propose changes, and schedule recurring tasks—all without exposing credentials.

  • Project
    Cmochance/codex-app-transfer

    A desktop gateway and configuration tool that allows the OpenAI Codex APP to use third-party AI models by translating its internal API requests into standard chat/completions protocols.