agent-orc

Runner — Development skill for Claude Code

Development community

Run coding-agent CLIs — Claude Code, Codex, Copilot, Gemini — from .NET: hardened spawning, streamed events, lifecycle & quota tracking.

How to install Runner

This entry records only its repository, not the path inside it, so there is no exact command to give. Open agent-orc/runner and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Runner does

Run coding-agent CLIs — Claude Code, Codex, Copilot, Gemini — from .NET: hardened spawning, streamed events, lifecycle & quota tracking.

Alternatives in Development

  • Understand Anything — Graphs that teach graphs that impress 80.7k ★
  • OpenAI Codex CLI — (Rust implementation) 67.8k ★
  • Vibe Kanban — Get 10X more out of Claude Code, Gemini CLI, Codex, Amp and other coding agents 23.5k ★

README

CodingAgentRunner

Run coding-agent CLIs — Claude Code, Codex, Antigravity, Gemini — from .NET: hardened spawning, streamed events, lifecycle, quota, metrics & optional rendering.

[![NuGet](https://img.shields.io/nuget/v/CodingAgentRunner.svg?label=NuGet)](https://www.nuget.org/packages/CodingAgentRunner) [![NuGet downloads](https://img.shields.io/nuget/dt/CodingAgentRunner.svg?label=downloads)](https://www.nuget.org/packages/CodingAgentRunner) [![CI](https://github.com/agent-orc/runner/actions/workflows/ci.yml/badge.svg)](https://github.com/agent-orc/runner/actions/workflows/ci.yml) [![License: Apache 2.0](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)

dotnet add package CodingAgentRunner            # core: spawning, events, lifecycle, quota, metrics
dotnet add package CodingAgentRunner.Rendering  # optional: Markdown/HTML rendering of agent output

Package pages: [CodingAgentRunner](https://www.nuget.org/packages/CodingAgentRunner) · [CodingAgentRunner.Rendering](https://www.nuget.org/packages/CodingAgentRunner.Rendering) · [GitHub releases](https://github.com/agent-orc/runner/releases)

**CodingAgentRunner** gives a .NET application an LLM on your local machine, using the coding-agent CLI you already sign in to — with no API keys. Run a single prompt, or a full multi-turn session. It launches and supervises terminal-native coding agents (Claude Code, OpenAI Codex, Google Antigravity / `agentapi`, and the legacy Gemini CLI) as child processes — reliably, especially on Windows.

It is the process-and-protocol layer for those CLIs: it spawns the agent CLI with the right binary, environment, and isolation; normalizes its `stream-json` output — a different frame dialect per CLI — into one structured event vocabulary; classifies the run's outcome; enforces a *platform-owns-git* boundary; tracks remaining quota with a cache that polls more often as usage approaches the limit; records run metrics; and can render agent Markdown throu