offline-mcp 0.1.0__tar.gz

offline_mcp-0.1.0/PKG-INFO

@@ -0,0 +1,43 @@
+Metadata-Version: 2.4
+Name: offline-mcp
+Version: 0.1.0
+Summary: Local AI inference for East Africa — Ollama, open weights, degraded-mode fallbacks. 6 tools. Thesis 
+License: MIT
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: fastmcp>=0.9
+
+# offline-mcp
+
+> Local AI inference infrastructure — Ollama wrapper, open weights directory, degraded-mode guide for East Africa.
+
+[![PyPI](https://img.shields.io/badge/PyPI-v0.1.0-blue?logo=pypi)](https://pypi.org/project/offline-mcp/)
+[![Thesis Layer](https://img.shields.io/badge/Thesis_Layer-L4_Offline_AI-red)](https://gabrielmahia.github.io/nairobi-stack)
+
+**Why:** Never assume OpenAI survives, Anthropic stays accessible, or export controls disappear.
+This matters more in Africa than anywhere else.
+
+**1st world equivalent:** Ollama, LLaMA, Mistral local deployment
+
+## Install
+```bash
+pip install offline-mcp
+```
+
+## Tools (6)
+| Tool | Description |
+|------|-------------|
+| `check_ollama_status` | Check if Ollama is running locally and list available models |
+| `run_local_inference` | Run a prompt through a local Ollama model |
+| `list_recommended_models` | Best open-weight models for East Africa use cases |
+| `degraded_mode_guide` | 4-level degraded mode architecture for offline operation |
+| `open_weights_directory` | Directory of open-weight models with Africa language support |
+| `local_deployment_guide` | Deployment guide for laptop, server, Raspberry Pi, Android |
+
+## Context
+Runs on a 50W solar panel + Raspberry Pi 4. Viable for rural Kenya clinics, schools, and community offices.
+
+→ [The Nairobi Stack](https://gabrielmahia.github.io/nairobi-stack)
+
+## License
+MIT © Gabriel Mahia | contact@aikungfu.dev

offline_mcp-0.1.0/README.md

@@ -0,0 +1,34 @@
+# offline-mcp
+
+> Local AI inference infrastructure — Ollama wrapper, open weights directory, degraded-mode guide for East Africa.
+
+[![PyPI](https://img.shields.io/badge/PyPI-v0.1.0-blue?logo=pypi)](https://pypi.org/project/offline-mcp/)
+[![Thesis Layer](https://img.shields.io/badge/Thesis_Layer-L4_Offline_AI-red)](https://gabrielmahia.github.io/nairobi-stack)
+
+**Why:** Never assume OpenAI survives, Anthropic stays accessible, or export controls disappear.
+This matters more in Africa than anywhere else.
+
+**1st world equivalent:** Ollama, LLaMA, Mistral local deployment
+
+## Install
+```bash
+pip install offline-mcp
+```
+
+## Tools (6)
+| Tool | Description |
+|------|-------------|
+| `check_ollama_status` | Check if Ollama is running locally and list available models |
+| `run_local_inference` | Run a prompt through a local Ollama model |
+| `list_recommended_models` | Best open-weight models for East Africa use cases |
+| `degraded_mode_guide` | 4-level degraded mode architecture for offline operation |
+| `open_weights_directory` | Directory of open-weight models with Africa language support |
+| `local_deployment_guide` | Deployment guide for laptop, server, Raspberry Pi, Android |
+
+## Context
+Runs on a 50W solar panel + Raspberry Pi 4. Viable for rural Kenya clinics, schools, and community offices.
+
+→ [The Nairobi Stack](https://gabrielmahia.github.io/nairobi-stack)
+
+## License
+MIT © Gabriel Mahia | contact@aikungfu.dev

offline_mcp-0.1.0/pyproject.toml

@@ -0,0 +1,18 @@
+[build-system]
+requires = ["setuptools>=68", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "offline-mcp"
+version = "0.1.0"
+description = "Local AI inference for East Africa — Ollama, open weights, degraded-mode fallbacks. 6 tools. Thesis "
+readme = "README.md"
+license = { text = "MIT" }
+requires-python = ">=3.10"
+dependencies = ["fastmcp>=0.9"]
+
+[project.scripts]
+offline-mcp = "offline_mcp.server:mcp.run"
+
+[tool.setuptools.packages.find]
+where = ["src"]

offline_mcp-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

offline_mcp-0.1.0/src/offline_mcp/__init__.py

@@ -0,0 +1,2 @@
+"""offline-mcp v0.1.0"""
+__version__ = "0.1.0"

offline_mcp-0.1.0/src/offline_mcp/server.py

@@ -0,0 +1,143 @@
+"""OfflineMCP — Local AI Inference Infrastructure (6 tools).
+Layer 4: Never assume OpenAI survives or stays accessible.
+Build for local inference, open weights, degraded operation.
+"""
+from __future__ import annotations
+from typing import Optional
+from fastmcp import FastMCP
+mcp = FastMCP(name="offline-mcp", description="Local AI inference infrastructure for Africa — Ollama, open weights, degraded-mode fallbacks. 6 tools.")
+
+RECOMMENDED_MODELS = [
+    {"name": "llama3.2:3b", "size_gb": 2.0, "ram_gb": 4, "use_case": "General civic queries, fast responses",
+     "command": "ollama pull llama3.2:3b", "africa_fit": "Excellent for low-RAM devices"},
+    {"name": "mistral:7b", "size_gb": 4.1, "ram_gb": 8, "use_case": "Better quality civic reasoning",
+     "command": "ollama pull mistral:7b", "africa_fit": "Good for higher-spec devices"},
+    {"name": "phi3:mini", "size_gb": 2.3, "ram_gb": 4, "use_case": "Microsoft's compact model, efficient",
+     "command": "ollama pull phi3:mini", "africa_fit": "Good balance of speed and quality"},
+    {"name": "gemma2:2b", "size_gb": 1.6, "ram_gb": 3, "use_case": "Google's compact model",
+     "command": "ollama pull gemma2:2b", "africa_fit": "Best for very low RAM environments"},
+    {"name": "qwen2.5:3b", "size_gb": 1.9, "ram_gb": 4, "use_case": "Multilingual including Swahili",
+     "command": "ollama pull qwen2.5:3b", "africa_fit": "Strong multilingual support for Swahili"},
+]
+
+@mcp.tool(name="check_ollama_status", description="Check if Ollama is running locally and list available models.")
+def check_ollama_status() -> dict:
+    import urllib.request as ur
+    try:
+        req = ur.Request("http://localhost:11434/api/tags", headers={"User-Agent": "offline-mcp/1.0"})
+        with ur.urlopen(req, timeout=3) as r:
+            d = __import__("json").loads(r.read())
+            models = [m["name"] for m in d.get("models", [])]
+            return {"ollama_running": True, "models_available": models,
+                    "count": len(models), "api_endpoint": "http://localhost:11434"}
+    except Exception as e:
+        return {"ollama_running": False, "error": str(e),
+                "install_guide": "curl https://ollama.ai/install.sh | sh",
+                "windows": "Download from ollama.ai/download",
+                "note": "Ollama provides local AI inference — no internet required once models are downloaded."}
+
+@mcp.tool(name="run_local_inference", description="Run a prompt through a local Ollama model.")
+def run_local_inference(prompt: str, model: Optional[str] = "llama3.2:3b") -> dict:
+    import urllib.request as ur
+    import json as js
+    data = js.dumps({"model": model, "prompt": prompt, "stream": False}).encode()
+    try:
+        req = ur.Request("http://localhost:11434/api/generate",
+                         data=data, method="POST",
+                         headers={"Content-Type":"application/json","User-Agent":"offline-mcp/1.0"})
+        with ur.urlopen(req, timeout=60) as r:
+            result = js.loads(r.read())
+            return {"status": "ok", "model": model, "response": result.get("response",""),
+                    "offline": True, "tokens": result.get("eval_count", 0)}
+    except Exception as e:
+        return {"status": "error", "error": str(e),
+                "fallback": "Ollama not running. Start with: ollama serve",
+                "offline": False}
+
+@mcp.tool(name="list_recommended_models", description="List recommended open-weight models for East Africa AI use cases.")
+def list_recommended_models(use_case: Optional[str] = None, max_ram_gb: Optional[int] = None) -> dict:
+    models = RECOMMENDED_MODELS
+    if max_ram_gb:
+        models = [m for m in models if m["ram_gb"] <= max_ram_gb]
+    if use_case:
+        u = use_case.lower()
+        models = [m for m in models if u in m["use_case"].lower() or u in m["africa_fit"].lower()] or models
+    return {"source": "Ollama model registry", "recommended_for_africa": models,
+            "install_commands": [m["command"] for m in models],
+            "note": "All models run locally — no API key, no internet after download, no cost per query."}
+
+@mcp.tool(name="degraded_mode_guide", description="Guide for operating AI systems when cloud connectivity fails.")
+def degraded_mode_guide() -> dict:
+    return {"source": "AI-KungFU offline architecture guide",
+            "why_this_matters": "Kenya's connectivity is improving but remains unreliable outside Nairobi/Mombasa. Power outages, data costs, and latency can make cloud AI unusable.",
+            "degraded_mode_levels": {
+                "level_1_full_online": "Cloud AI + all MCP servers. Best quality. Requires stable internet.",
+                "level_2_local_llm": "Ollama local model + MCP servers. Good quality. Requires local setup.",
+                "level_3_cached_responses": "Pre-cached MCP responses + local model. Acceptable quality. Fully offline.",
+                "level_4_static_reference": "Static lookup tables and reference data only. No AI inference.",
+            },
+            "implementation": {
+                "step_1": "Install Ollama: curl https://ollama.ai/install.sh | sh",
+                "step_2": "Download a model: ollama pull llama3.2:3b (2GB)",
+                "step_3": "Test: ollama run llama3.2:3b 'Hello, what is the KRA PIN number format?'",
+                "step_4": "Configure MCP clients to use localhost:11434 as fallback",
+            },
+            "hardware_minimums": {"ram_gb": 4, "storage_gb": 5, "cpu": "Any modern CPU (no GPU required for 3B models)"},
+            "raspberry_pi": "Ollama runs on Raspberry Pi 4 (4GB RAM) — viable for rural offline deployments."}
+
+@mcp.tool(name="open_weights_directory", description="Directory of open-weight AI models suitable for East Africa civic use cases.")
+def open_weights_directory(use_case: Optional[str] = None) -> dict:
+    MODELS = [
+        {"name": "Llama 3.2 (Meta)", "sizes": ["1B", "3B", "11B"], "languages": ["en", "sw (limited)"],
+         "licence": "Llama 3.2 Community License (free for most uses)", "source": "ollama pull llama3.2"},
+        {"name": "Mistral 7B (Mistral AI)", "sizes": ["7B"], "languages": ["en", "fr"],
+         "licence": "Apache 2.0 — completely free and open", "source": "ollama pull mistral"},
+        {"name": "Gemma 2 (Google)", "sizes": ["2B", "9B", "27B"], "languages": ["en", "multilingual"],
+         "licence": "Gemma Terms of Use (free for most uses)", "source": "ollama pull gemma2:2b"},
+        {"name": "Qwen 2.5 (Alibaba)", "sizes": ["0.5B", "1.5B", "3B", "7B"], "languages": ["en", "zh", "sw (good)"],
+         "licence": "Apache 2.0", "source": "ollama pull qwen2.5:3b"},
+        {"name": "Aya Expanse (Cohere for AI)", "sizes": ["8B", "32B"], "languages": ["101 languages including sw"],
+         "licence": "CC BY-NC 4.0", "source": "huggingface.co/CohereForAI/aya-expanse-8b"},
+        {"name": "AfroLM / AfriBERTa (Masakhane)", "sizes": ["Various"], "languages": ["African languages"],
+         "licence": "Apache 2.0", "source": "huggingface.co/masakhane"},
+    ]
+    if use_case:
+        u = use_case.lower()
+        MODELS = [m for m in MODELS if u in str(m).lower()] or MODELS
+    return {"source": "HuggingFace, Ollama, Masakhane", "open_models": MODELS,
+            "africa_note": "Aya Expanse and Masakhane models have best African language support.",
+            "deployment": "All models above can be deployed locally via Ollama or llama.cpp."}
+
+@mcp.tool(name="local_deployment_guide", description="Guide to deploying local AI inference on modest hardware in Kenya/East Africa.")
+def local_deployment_guide(device_type: Optional[str] = "laptop") -> dict:
+    GUIDES = {
+        "laptop": {
+            "recommended_model": "llama3.2:3b or gemma2:2b",
+            "min_specs": "4GB RAM, 5GB storage, any modern CPU",
+            "setup": "1. Install Ollama (ollama.ai) 2. ollama pull llama3.2:3b 3. ollama serve",
+            "inference_speed": "~5-20 tokens/second on 4-core CPU",
+        },
+        "server": {
+            "recommended_model": "mistral:7b or llama3.1:8b",
+            "min_specs": "16GB RAM, 20GB storage, 4+ core CPU",
+            "setup": "Docker: docker run -p 11434:11434 ollama/ollama",
+            "inference_speed": "~20-50 tokens/second",
+        },
+        "raspberry_pi": {
+            "recommended_model": "gemma2:2b or llama3.2:1b",
+            "min_specs": "Raspberry Pi 4 with 4GB RAM, 16GB microSD",
+            "setup": "curl https://ollama.ai/install.sh | sh && ollama pull gemma2:2b",
+            "inference_speed": "~2-5 tokens/second (usable for simple queries)",
+            "power": "~15W — solar-compatible for off-grid deployments",
+        },
+        "android": {
+            "recommended_model": "llama3.2:1b via llama.cpp",
+            "app": "PocketPal AI (open source) or MLC Chat",
+            "min_specs": "Android phone with 6GB+ RAM",
+            "note": "Experimental but viable for very simple queries.",
+        },
+    }
+    guide = GUIDES.get(device_type.lower(), GUIDES["laptop"])
+    return {"source": "AI-KungFU offline deployment guide", "device_type": device_type,
+            **guide, "all_options": list(GUIDES.keys()),
+            "solar_note": "Ollama on Raspberry Pi can run on a 50W solar panel — viable for rural Kenya clinics, schools, and offices."}

offline_mcp-0.1.0/src/offline_mcp.egg-info/PKG-INFO

@@ -0,0 +1,43 @@
+Metadata-Version: 2.4
+Name: offline-mcp
+Version: 0.1.0
+Summary: Local AI inference for East Africa — Ollama, open weights, degraded-mode fallbacks. 6 tools. Thesis 
+License: MIT
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: fastmcp>=0.9
+
+# offline-mcp
+
+> Local AI inference infrastructure — Ollama wrapper, open weights directory, degraded-mode guide for East Africa.
+
+[![PyPI](https://img.shields.io/badge/PyPI-v0.1.0-blue?logo=pypi)](https://pypi.org/project/offline-mcp/)
+[![Thesis Layer](https://img.shields.io/badge/Thesis_Layer-L4_Offline_AI-red)](https://gabrielmahia.github.io/nairobi-stack)
+
+**Why:** Never assume OpenAI survives, Anthropic stays accessible, or export controls disappear.
+This matters more in Africa than anywhere else.
+
+**1st world equivalent:** Ollama, LLaMA, Mistral local deployment
+
+## Install
+```bash
+pip install offline-mcp
+```
+
+## Tools (6)
+| Tool | Description |
+|------|-------------|
+| `check_ollama_status` | Check if Ollama is running locally and list available models |
+| `run_local_inference` | Run a prompt through a local Ollama model |
+| `list_recommended_models` | Best open-weight models for East Africa use cases |
+| `degraded_mode_guide` | 4-level degraded mode architecture for offline operation |
+| `open_weights_directory` | Directory of open-weight models with Africa language support |
+| `local_deployment_guide` | Deployment guide for laptop, server, Raspberry Pi, Android |
+
+## Context
+Runs on a 50W solar panel + Raspberry Pi 4. Viable for rural Kenya clinics, schools, and community offices.
+
+→ [The Nairobi Stack](https://gabrielmahia.github.io/nairobi-stack)
+
+## License
+MIT © Gabriel Mahia | contact@aikungfu.dev

offline_mcp-0.1.0/src/offline_mcp.egg-info/SOURCES.txt

@@ -0,0 +1,10 @@
+README.md
+pyproject.toml
+src/offline_mcp/__init__.py
+src/offline_mcp/server.py
+src/offline_mcp.egg-info/PKG-INFO
+src/offline_mcp.egg-info/SOURCES.txt
+src/offline_mcp.egg-info/dependency_links.txt
+src/offline_mcp.egg-info/entry_points.txt
+src/offline_mcp.egg-info/requires.txt
+src/offline_mcp.egg-info/top_level.txt

offline_mcp-0.1.0/src/offline_mcp.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

offline_mcp-0.1.0/src/offline_mcp.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+offline-mcp = offline_mcp.server:mcp.run

offline_mcp-0.1.0/src/offline_mcp.egg-info/requires.txt

@@ -0,0 +1 @@
+fastmcp>=0.9

offline_mcp-0.1.0/src/offline_mcp.egg-info/top_level.txt

@@ -0,0 +1 @@
+offline_mcp