@tecet/ollm 0.1.4 → 0.1.5

package/docs/LLM Models/LLM_GettingStarted.md

@@ -1,748 +0,0 @@
-# Getting Started with Model Management
-
-**Quick Start Guide**
-
-This guide will help you get started with model management, routing, memory, templates, and project profiles in OLLM CLI.
-
----
-
-## 📋 Table of Contents
-
-1. [Introduction](#introduction)
-2. [Prerequisites](#prerequisites)
-3. [Quick Start](#quick-start)
-4. [Basic Model Management](#basic-model-management)
-5. [Using Model Routing](#using-model-routing)
-6. [Working with Memory](#working-with-memory)
-7. [Using Templates](#using-templates)
-8. [Project Profiles](#project-profiles)
-9. [Next Steps](#next-steps)
-
-**See Also:**
-
-- [Model Management Overview](3%20projects/OLLM%20CLI/LLM%20Models/README.md)
-- [Model Commands](Models_commands.md)
-- [Configuration Guide](Models_configuration.md)
-
----
-
-## Introduction
-
-OLLM CLI provides comprehensive model management capabilities that help you:
-
-- **Manage models**: List, download, delete, and inspect models
-- **Route intelligently**: Automatically select appropriate models for tasks
-- **Remember context**: Store facts and preferences across sessions
-- **Use templates**: Create reusable prompts with variables
-- **Configure projects**: Auto-detect and apply project-specific settings
-
-This guide covers the basics to get you started quickly.
-
----
-
-## Prerequisites
-
-**Required:**
-
-- OLLM CLI installed and configured
-- Ollama (or compatible provider) running
-- At least one model installed
-
-**Optional:**
-
-- Project workspace for profile detection
-- Custom templates directory
-
-**Check your setup:**
-
-```bash
-# Verify OLLM CLI is installed
-ollm --version
-
-# Check provider connection
-/model list
-```
-
----
-
-## Quick Start
-
-### 1. List Available Models
-
-See what models you have installed:
-
-```bash
-/model list
-```
-
-**Output:**
-
-```
-Available Models:
-  ● llama3.1:8b (loaded)     4.7 GB   Modified 2 days ago
-    mistral:7b               4.1 GB   Modified 1 week ago
-    codellama:7b             3.8 GB   Modified 3 days ago
-```
-
-### 2. Download a Model
-
-Pull a new model from the registry:
-
-```bash
-/model pull llama3.1:8b
-```
-
-**Progress display:**
-
-```
-Pulling llama3.1:8b...
-[████████████████████████] 100% 4.7 GB @ 12.3 MB/s
-Model llama3.1:8b ready.
-```
-
-### 3. View Model Details
-
-Get detailed information about a model:
-
-```bash
-/model info llama3.1:8b
-```
-
-**Output:**
-
-```
-Model: llama3.1:8b
-Size: 4.7 GB
-Parameters: 8B
-Context Length: 128,000 tokens
-Quantization: Q4_K_M
-Family: llama
-Capabilities:
-  ✓ Tool calling
-  ✓ Streaming
-  ✗ Vision
-```
-
-### 4. Use a Model
-
-Switch to a specific model:
-
-```bash
-/model use llama3.1:8b
-```
-
----
-
-## Basic Model Management
-
-### Listing Models
-
-View all available models with details:
-
-```bash
-/model list
-```
-
-**Shows:**
-
-- Model name and version
-- Size on disk
-- Last modified date
-- Load status (loaded/unloaded)
-
-### Pulling Models
-
-Download models from the provider registry:
-
-```bash
-# Pull a specific model
-/model pull llama3.1:8b
-
-# Pull with tag
-/model pull mistral:7b-instruct
-```
-
-**Features:**
-
-- Real-time progress display
-- Transfer rate monitoring
-- Cancellable with Ctrl+C
-
-### Deleting Models
-
-Remove models to free disk space:
-
-```bash
-/model delete codellama:7b
-```
-
-**Safety features:**
-
-- Confirmation prompt
-- Shows space to be freed
-- Automatic unload if currently loaded
-
-### Keeping Models Loaded
-
-Keep frequently-used models in memory for faster responses:
-
-```bash
-# Keep a model loaded
-/model keep llama3.1:8b
-
-# Unload a model
-/model unload llama3.1:8b
-```
-
-**Benefits:**
-
-- Eliminates model load time (2-5 seconds)
-- Reduces latency for subsequent requests
-- Useful for interactive sessions
-
-### Unknown Model Handling
-
-When you switch to a model that isn't in the system's database, OLLM CLI will prompt you to configure its tool support:
-
-```bash
-/model use custom-model:latest
-```
-
-**Prompt:**
-
-```
-Unknown model detected: custom-model:latest
-Does this model support function calling (tools)?
-  [y] Yes, it supports tools
-  [n] No, it doesn't support tools
-  [a] Auto-detect (test with a sample request)
-```
-
-**Options:**
-
-1. **Yes (y)**: Manually confirm the model supports tools
-   - Tools will be enabled for this model
-   - Choice is saved to `~/.ollm/user_models.json`
-
-2. **No (n)**: Manually confirm the model doesn't support tools
-   - Tools will be disabled for this model
-   - Choice is saved to `~/.ollm/user_models.json`
-
-3. **Auto-detect (a)**: Let the system test the model
-   - Sends a test request with a minimal tool schema
-   - Detects whether the model accepts or rejects tools
-   - Takes ~5 seconds with timeout
-   - Result is saved automatically
-
-**Timeout:** If you don't respond within 30 seconds, the system defaults to tools disabled (safe default).
-
-**Why this matters:**
-
-- Sending tools to models that don't support them causes errors
-- Proper configuration ensures smooth operation
-- Your choice is remembered for future sessions
-
-### Auto-Detect Details
-
-When you choose auto-detect, here's what happens:
-
-**Process:**
-
-1. System sends a test request to the model with a minimal tool schema
-2. Monitors the response for tool-related errors
-3. Times out after 5 seconds if no response
-4. Saves the result to `~/.ollm/user_models.json`
-
-**Success indicators:**
-
-- Model accepts the request without tool-related errors
-- Tool support is enabled and saved
-
-**Failure indicators:**
-
-- Model returns errors like "unknown field: tools"
-- Model returns 400 status with tool-related error messages
-- Tool support is disabled and saved
-
-**Fallback:**
-
-- If auto-detect fails or times out, tools are disabled (safe default)
-- You can manually update the setting later
-
-**System messages:**
-
-```
-Auto-detecting tool support for custom-model:latest...
-Tool support detected: Enabled
-```
-
-or
-
-```
-Auto-detecting tool support for custom-model:latest...
-Tool support detected: Disabled
-```
-
----
-
-## Using Model Routing
-
-### What is Model Routing?
-
-Model routing automatically selects appropriate models based on task type, eliminating the need for manual model selection.
-
-### Enable Routing
-
-Add to your configuration (`~/.ollm/config.yaml`):
-
-```yaml
-model:
-  routing:
-    enabled: true
-    defaultProfile: general
-```
-
-### Routing Profiles
-
-Four built-in profiles optimize for different use cases:
-
-**Fast Profile** - Quick responses with smaller models:
-
-```yaml
-routing:
-  defaultProfile: fast
-```
-
-**General Profile** - Balanced performance for most tasks:
-
-```yaml
-routing:
-  defaultProfile: general
-```
-
-**Code Profile** - Optimized for code generation:
-
-```yaml
-routing:
-  defaultProfile: code
-```
-
-**Creative Profile** - Creative writing and storytelling:
-
-```yaml
-routing:
-  defaultProfile: creative
-```
-
-### Profile Overrides
-
-Specify models for specific profiles:
-
-```yaml
-model:
-  routing:
-    enabled: true
-    defaultProfile: general
-    overrides:
-      code: deepseek-coder:6.7b
-      fast: phi3:mini
-```
-
-### Manual Override
-
-You can always manually select a model:
-
-```bash
-/model use llama3.1:8b
-```
-
-This overrides routing for the current session.
-
----
-
-## Working with Memory
-
-### What is Memory?
-
-The memory system stores facts, preferences, and context across sessions, so you don't have to repeat information.
-
-### Adding Memories
-
-Store information for future sessions:
-
-```bash
-# Add a simple memory
-/memory add user_name Alice
-
-# Add with category
-/memory add preferred_language TypeScript --category preference
-```
-
-### Listing Memories
-
-View all stored memories:
-
-```bash
-/memory list
-```
-
-**Output:**
-
-```
-Stored Memories:
-  user_name: Alice (preference)
-  preferred_language: TypeScript (preference)
-  project_type: monorepo (context)
-```
-
-### Searching Memories
-
-Find specific memories:
-
-```bash
-/memory search project
-```
-
-### Forgetting Memories
-
-Remove memories you no longer need:
-
-```bash
-# Forget a specific memory
-/memory forget old_preference
-
-# Clear all memories
-/memory clear
-```
-
-### LLM-Initiated Memory
-
-The LLM can also store memories during conversation:
-
-```
-User: My name is Alice and I prefer TypeScript.
-Assistant: I'll remember that. [Stores: user_name=Alice, preferred_language=TypeScript]
-```
-
-### How Memory Works
-
-1. Memories are stored in `~/.ollm/memory.json`
-2. At session start, memories are loaded
-3. Memories are injected into the system prompt (within token budget)
-4. Recently accessed memories are prioritized
-5. Access count and timestamps are tracked
-
----
-
-## Using Templates
-
-### What are Templates?
-
-Templates are reusable prompts with variable substitution, allowing you to quickly use common prompts with different inputs.
-
-### Listing Templates
-
-View available templates:
-
-```bash
-/template list
-```
-
-**Output:**
-
-```
-Available Templates:
-  code_review - Review code for quality and security
-  explain_code - Explain how code works
-  write_tests - Generate unit tests for code
-```
-
-### Using Templates
-
-Apply a template with variables:
-
-```bash
-# Use a template with variables
-/template use code_review language=TypeScript code="function add(a, b) { return a + b; }"
-
-# Variables with spaces
-/template use explain_code language="Python" code="def factorial(n): return 1 if n <= 1 else n * factorial(n-1)"
-```
-
-### Creating Templates
-
-Create a new template:
-
-```bash
-/template create my_template
-```
-
-**Template format (YAML):**
-
-```yaml
-name: code_review
-description: Review code for quality and security
-template: "Review this {language} code for {focus:bugs and security}:\n\n{code}"
-variables:
-  - name: language
-    required: true
-    description: Programming language
-  - name: focus
-    required: false
-    default: 'bugs and security'
-    description: Review focus areas
-  - name: code
-    required: true
-    description: Code to review
-```
-
-### Template Locations
-
-Templates are loaded from:
-
-- User templates: `~/.ollm/templates/`
-- Workspace templates: `.ollm/templates/`
-
-Workspace templates override user templates with the same name.
-
----
-
-## Project Profiles
-
-### What are Project Profiles?
-
-Project profiles auto-detect your project type and apply appropriate settings automatically.
-
-### Auto-Detection
-
-OLLM CLI detects project type from characteristic files:
-
-```bash
-/project detect
-```
-
-**Detected types:**
-
-- **TypeScript**: `package.json` with TypeScript dependencies
-- **Python**: `requirements.txt`, `pyproject.toml`, `setup.py`
-- **Rust**: `Cargo.toml`
-- **Go**: `go.mod`
-
-### Using a Profile
-
-Manually select a profile:
-
-```bash
-/project use typescript
-```
-
-### Initializing a Project
-
-Create a project configuration file:
-
-```bash
-/project init
-```
-
-This creates `.ollm/project.yaml` with the selected profile settings.
-
-### Project Configuration
-
-Example `.ollm/project.yaml`:
-
-```yaml
-# Project profile
-profile: typescript
-
-# Override global model
-model: deepseek-coder:6.7b
-
-# Project-specific routing
-routing:
-  defaultProfile: code
-
-# Project-specific options
-options:
-  temperature: 0.3
-  maxTokens: 4096
-```
-
-### Built-in Profiles
-
-**TypeScript Profile:**
-
-- Code-optimized model
-- Code routing profile
-- File and shell tools enabled
-
-**Python Profile:**
-
-- Code-optimized model
-- Code routing profile
-- Python-specific tools
-
-**Rust Profile:**
-
-- Code-optimized model
-- Emphasis on memory safety
-
-**Go Profile:**
-
-- Code-optimized model
-- Emphasis on concurrency
-
-**Documentation Profile:**
-
-- Writing-optimized model
-- Creative routing profile
-
----
-
-## Managing Tools
-
-### What are Tools?
-
-Tools are functions that the LLM can call to perform actions like reading files, executing shell commands, searching the web, and more. OLLM CLI provides 15 built-in tools organized into 6 categories.
-
-### Tools Panel
-
-Access the Tools Panel to enable or disable individual tools:
-
-**Navigation:**
-
-- Switch to the Tools tab in the UI
-- Use keyboard shortcuts to navigate:
-  - `↑/↓`: Navigate between tools
-  - `←/→/Enter`: Toggle tool on/off
-  - `Tab`: Switch between tabs
-
-**Tool Categories:**
-
-1. **File Operations** (4 tools)
-   - `fsWrite`: Create or overwrite files
-   - `fsAppend`: Append content to files
-   - `strReplace`: Replace text in files
-   - `deleteFile`: Delete files
-
-2. **File Discovery** (5 tools)
-   - `readFile`: Read file contents
-   - `readMultipleFiles`: Read multiple files at once
-   - `listDirectory`: List directory contents
-   - `fileSearch`: Search for files by name
-   - `grepSearch`: Search file contents with regex
-
-3. **Shell** (4 tools)
-   - `executePwsh`: Execute shell commands
-   - `controlPwshProcess`: Manage background processes
-   - `listProcesses`: List running processes
-   - `getProcessOutput`: Read process output
-
-4. **Web** (2 tools)
-   - `remote_web_search`: Search the web
-   - `webFetch`: Fetch content from URLs
-
-5. **Memory** (1 tool)
-   - `userInput`: Get input from the user
-
-6. **Context** (4 tools)
-   - `prework`: Acceptance criteria testing prework
-   - `taskStatus`: Update task status
-   - `updatePBTStatus`: Update property-based test status
-   - `invokeSubAgent`: Delegate to specialized agents
-
-### Enabling/Disabling Tools
-
-**Why disable tools?**
-
-- Reduce the number of tools sent to the LLM (improves focus)
-- Prevent certain actions (e.g., disable shell execution for safety)
-- Customize tool availability per project
-
-**How to disable:**
-
-1. Navigate to the Tools tab
-2. Use arrow keys to select a tool
-3. Press Enter or Left/Right to toggle
-
-**Visual indicators:**
-
-- `[✓]` Tool is enabled
-- `[ ]` Tool is disabled
-
-**Persistence:**
-
-- Tool settings are saved to `~/.ollm/settings.json`
-- Settings persist across sessions
-- Workspace-specific settings can override user settings
-
-### Tool Filtering
-
-Tools are filtered in two stages:
-
-1. **Model Capability Check**: If the current model doesn't support function calling, all tools are automatically disabled
-2. **User Preference Check**: Even if the model supports tools, you can disable specific tools via the Tools Panel
-
-**System message when tools are disabled:**
-
-```
-Switched to gemma3:1b. Tools: Disabled
-```
-
-### Model Tool Support
-
-Some models don't support function calling. When you switch to such a model:
-
-- Tools are automatically disabled
-- System prompt includes a note: "This model does not support function calling"
-- Tools Panel shows: "Model doesn't support tools"
-- You can still view and configure tool preferences for when you switch back to a tool-capable model
-
----
-
-## Next Steps
-
-### Learn More
-
-**Model Management:**
-
-- [Model Commands Reference](Models_commands.md)
-- [Model Architecture](Models_architecture.md)
-- [Configuration Guide](Models_configuration.md)
-
-**Routing:**
-
-- [Routing User Guide](3%20projects/OLLM%20CLI/LLM%20Models/routing/user-guide.md)
-- [Routing Development Guide](3%20projects/OLLM%20CLI/LLM%20Models/routing/development-guide.md)
-- [Profiles Reference](profiles-reference.md)
-
-**Memory:**
-
-- [Memory User Guide](3%20projects/OLLM%20CLI/LLM%20Models/memory/user-guide.md)
-- [Memory API Reference](api-reference.md)
-
-**Templates:**
-
-- [Templates User Guide](3%20projects/OLLM%20CLI/LLM%20Models/templates/user-guide.md)
-- [Template Reference](template-reference.md)
-
-**Profiles:**
-
-- [Profiles User Guide](3%20projects/OLLM%20CLI/LLM%20Models/profiles/user-guide.md)
-- [Built-in Profiles](built-in-profiles.md)
-
-### Advanced Topics
-
-- [Custom Routing Profiles](3%20projects/OLLM%20CLI/LLM%20Models/routing/development-guide.md)
-- [Template Libraries](template-reference.md)
-- [API Reference](api/)
-
-### Get Help
-
-- [Troubleshooting Guide](../Troubleshooting.md)
-- GitHub Issues (https://github.com/ollm/ollm-cli/issues)
-- [Community Forum](#)
-
----
-
-**Last Updated:** 2026-01-16  
-**Version:** 0.1.0