I build local-first AI developer tools -- utilities that make LLMs useful in real workflows without cloud lock-in or API costs.
Local-LLM-Tools, Open Source, made in the Ruhrgebiet. Fokus: lauffaehige LLM-Workflows ohne Cloud-Abhaengigkeit.
| Project | What it does | Install |
|---|---|---|
| ollama-commit | AI-generated Git commit messages via local Ollama | pip install ollama-commit |
| ollama-sync | Sync Ollama models and tags between multiple hosts (diff/push/pull/mirror via SSH+tar) | pip install git+https://github.com/dennisreichenberg/ollama-sync |
| ollama-tui | Keyboard-driven Terminal UI for chatting with local Ollama models | pip install git+https://github.com/dennisreichenberg/ollama-tui |
| devlog | Log daily work, generate standups & PR descriptions via local LLM | pip install devlog |
| llm-bench | CLI benchmark suite for local LLM models | pip install llm-bench |
| llm-shell | AI-powered shell command suggester via plain-text descriptions | pip install llm-shell |
| local-rag | Private document Q&A with local LLMs | pip install ollama-local-rag |
| model-manager | Unified Ollama model admin -- list, pull, remove, tag | pip install model-manager |
| prompt-lab | Side-by-side prompt testing across local Ollama models | pip install git+https://github.com/dennisreichenberg/prompt-lab |
| Project | What it does | Install |
|---|---|---|
| vllm-bench | CLI benchmark for vLLM server performance (TTFT, throughput, ITL) | pip install git+https://github.com/dennisreichenberg/vllm-bench |
| vllm-config | Hardware-aware vLLM configuration optimizer | pip install git+https://github.com/dennisreichenberg/vllm-config |
| vllm-monitor | Real-time terminal dashboard for vLLM server metrics | pip install git+https://github.com/dennisreichenberg/vllm-monitor |
| Project | What it does | Install |
|---|---|---|
| ai-stack | Lifecycle supervisor for the local LLM stack -- start, stop, restart and status for Ollama, vLLM, ollama-proxy/llm-gateway, n8n | pip install git+https://github.com/dennisreichenberg/ai-stack |
| llm-cache | Semantic response cache in front of OpenAI-compatible LLM gateways (ollama-proxy, llm-gateway, vanilla Ollama) | pip install git+https://github.com/dennisreichenberg/llm-cache |
| llm-usage | Historical request and cost accounting -- tracks tokens, latency and cloud costs across Ollama, vLLM and cloud API backends | pip install git+https://github.com/dennisreichenberg/llm-usage |
| ollama-proxy | Unified OpenAI-compatible API router for Ollama and vLLM backends | pip install ollama-proxy |
| agent-runner | CLI agent loop framework with tool use for local LLMs | pip install agent-runner |
| model-eval | Evaluate local LLM quality -- exact-match, fuzzy, and LLM-as-judge scoring | pip install git+https://github.com/dennisreichenberg/model-eval |
| rag-eval | Evaluate local RAG pipelines -- recall@k, MRR, BLEU/ROUGE and LLM-as-judge | pip install rag-eval |
External open-source tools that Dennis self-hosts alongside his CLI tools. Configs, secrets templates and integration notes live in reichenberg-ruhr. These are third-party OSS projects, not authored by Dennis.
| Component | What it does | Dennis's Config |
|---|---|---|
| Langfuse | Self-hosted LLM observability -- traces, evals, prompt management and cost tracking across all local-LLM tools | config (UI on :3030) |
| LiteLLM Proxy | OpenAI-compatible unified API gateway for Ollama and vLLM, with optional Langfuse callback | config (proxy on :4000) |
All my tools run 100% locally. No API keys. No usage costs. No data leaving your machine. If you have Ollama installed, you already have everything you need.
🌐 reichenberg.ruhr · 📦 PyPI
- agent-runner -- CLI-based local agent loop with tool-use via Ollama
- ai-stack -- Lifecycle supervisor for the local AI stack (Ollama, vLLM, ollama-proxy/llm-gateway, n8n)
- devlog -- AI-powered developer diary: log your work, generate weekly summaries and PR descriptions via local LLMs
- llm-bench -- Benchmark local LLMs via Ollama -- measure tokens/s, TTFT, and memory usage
- llm-cache -- Semantic response cache in front of OpenAI-compatible LLM gateways (ollama-proxy, llm-gateway, vanilla Ollama)
- llm-usage -- Historical request and cost accounting across all LLM backends (Ollama/vLLM + cloud)
- ollama-sync -- Sync Ollama models and tags between multiple hosts (diff/push/pull/mirror via SSH+tar)