Harvey Reference Manual

This document provides a comprehensive reference for all Harvey commands, configuration options, and platform-specific information.

📖 Command Reference

All Harvey commands are organized into logical categories. Each command has its own man page with detailed usage information.

Core Commands

• harvey.1.md — Primary man page: name, synopsis, description, options, environment variables, and all available slash commands
• harvey.7.md — General concepts and overview of Harvey’s design philosophy

File Operations

• harvey-read.7.md — Read file contents into the conversation context
• harvey-write.7.md — Write assistant output to a file
• harvey-attach.7.md — Attach a file (image, PDF, or text) to the next turn
• harvey-files.7.md — List directory contents in the workspace
• harvey-file-tree.7.md — Display a recursive directory tree
• harvey-read-dir.7.md — Read all eligible files in a directory tree into context
• harvey-read-pdf.7.md — Extract text from a PDF using poppler utilities
• harvey-editing.7.md — Edit files using the configured editor
• harvey-format.7.md — Format workspace source files in-place

Context and Search

• harvey-search.7.md — Regex search across workspace files
• harvey-context.7.md — Manage pinned context that survives /clear
• harvey-git.7.md — Read-only git commands (status, diff, log, show, blame)

RAG (Retrieval-Augmented Generation)

• harvey-rag.7.md — Manage retrieval-augmented generation stores
• harvey-learn.7.md — Learn from context and build knowledge
• harvey-inspect.7.md — Inspect model capabilities and metadata

Session Management

• harvey-session.7.md — Session management commands
• harvey-record.7.md — Start, stop, and manage session recordings
• harvey-rename.7.md — Rename the active session file
• harvey-summarize.7.md — Condense conversation history to free context window space
• harvey-clear.7.md — Reset conversation history
• harvey-loop.7.md — Manage the REPL loop

Memory System

• harvey-memory.7.md — Manage the session-experience memory store

Knowledge Base

• harvey-kb.7.md — Query and update the SQLite knowledge base

Ollama Integration

• harvey-ollama.7.md — Manage the local Ollama server and installed models
• harvey-model-alias.7.md — Create and manage model aliases

Routing

• harvey-routing.7.md — Manage named remote LLM endpoints for @mention routing

Skills

• harvey-skills.7.md — Discover and load SKILL.md-format skills into context
• harvey-skill-set.7.md — Load and manage named bundles of skills

Pipelines

• harvey-pipeline.7.md — Chain Markdown prompt files as discrete steps

Security

• harvey-security.7.md — Unified security posture overview

Execution

• harvey-run.7.md — Run shell commands safely within the workspace

System Status

• harvey-status.7.md — Show active backend, token usage, routing, and debug state
• harvey-hint.7.md — Show actionable suggestions for improving results (RAG, memory, KB)

🔧 Configuration Reference

Main Configuration

• CONFIGURATION.md — Complete workspace configuration reference for agents/harvey.yaml, agents/routes.json, and environment variables

Workspace Configuration Files

• agents/harvey.yaml — Main configuration file
• agents/routes.json — Named route definitions for multi-backend support
• agents/knowledge.db — SQLite knowledge base
• HARVEY.md — System prompt for the workspace

Environment Variables

See harvey.1.md for complete list: - ANTHROPIC_API_KEY — API key for Anthropic Claude - DEEPSEEK_API_KEY — API key for DeepSeek - GEMINI_API_KEY — API key for Google Gemini (also accepts GOOGLE_API_KEY) - MISTRAL_API_KEY — API key for Mistral - OPENAI_API_KEY — API key for OpenAI

Note: All API key variables are filtered from child process environments

🌍 Platform-Specific Information

Installation Notes

• INSTALL.md — Main installation guide for all platforms
• INSTALL_NOTES_macOS.md — macOS-specific installation notes and requirements
• INSTALL_NOTES_Windows.md — Windows-specific installation notes and requirements

Supported Architectures

• Linux: x86_64, aarch64, armv7l
• macOS: x86_64, arm64
• Windows: x86_64

📚 Additional Reference

Models

• models.md — Models in use and evaluation status
• model_guide.md — Model selection guide based on /ollama probe results
• Llamafile_notes.md — Mozilla AI’s single-file runnable models (note: not currently integrated with Harvey)

Development

• TESTING.md — Running tests, test architecture, and CI/CD setup
• CHANGES.md — Version history and release notes
• TODO.md — Current action items, known bugs, and planned features
• further_reading.md — Additional reading and resources

Formatting

• page.tmpl — Template for HTML page generation

🎯 Quick Command Lookup

Most Common Commands

Model Management

RAG Operations

Knowledge Base

Security