tinycmdr logo

tinycmdr

Free

tinycmdr: the high-efficiency agent harness for local & self-hosted LLMs (llama.cpp, vLLM, Ollama, SGLang): ~4.1K token overhead, stable prefix caching, built-in ops guards, zero infrastructure.

Visit Website
Related by shared tags and keywordsMingbird AgentqwenCerveau
ai-agentagent-harnessclillamacpplocal-llmaiai assistantautomation

Overview

tinycmdr is an autonomous ops agent for self-hosted models. It runs on the machine it manages, accepts tasks, and performs work using real tools such as shell, execute_code, edit_file, and fetch_url. The agent narrates its actions and keeps its state in plain files beside itself. It is designed for small and local models, with a stable, cache-friendly prompt prefix and a runtime intended to prevent a confused model from wedging the inference slot. It can be reached from a terminal or chat platforms including Mattermost and Telegram, using one process and the same conversations. Installation is available for Windows, Linux, and macOS, requiring Python 3.10–3.12. The agent supports commands for setup, status, doctor, health, model selection, logs, version, process state, update, and restart. It can be extended with prose skills placed in a skills folder and custom tools placed in a tools folder.

Pricing

Free

Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.

Visit Website

Prices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.

Use Cases

  • Teams use tinycmdr to run an autonomous ops agent on the machine it manages.
  • Teams use tinycmdr to execute tasks with real tools such as shell, execute_code, edit_file, and fetch_url.
  • Teams use tinycmdr to interact with the agent from a terminal or chat platforms like Mattermost and Telegram.
  • Teams use tinycmdr to check the installation and identify problems with the doctor command.
  • Teams use tinycmdr to get a one-line health check with an exit code for scripts, without network access.
  • Teams use tinycmdr to select a model from a list, with a wizard offered if the endpoint is dead.
  • Teams use tinycmdr to extend the agent with prose skills by dropping a runbook folder into skills/.
  • Teams use tinycmdr to add custom tools by dropping a .py file into tools/.

Features

Agent harness for local LLMs

Support for self-hosted LLMs

Works with llama.cpp and vLLM

Works with Ollama and SGLang

Approximately 4.1K tokens of overhead

Stable prefix caching for agents

Built-in operational safeguards

Runs without additional infrastructure

Reviews

No reviews yet. Be the first to share your experience!