Wilk provides a comprehensive command-line interface with 80+ commands for agent management, conversation handling, memory management, and system administration. All commands follow consistent patterns adapted from LibreChat's proven interaction models.
wilk [global-options] <command> [subcommand] [options] [arguments]--config <path> # Specify custom config file
--project <path> # Set project directory
--profile <name> # Use specific profile
--verbose, -v # Verbose output
--quiet, -q # Suppress non-essential output
--no-thinking # Disable agent thinking display
--show-thinking # Force show agent thinking
--thinking-live # Show live thinking stream
--dry-run # Show what would be done without executing
--format <json|yaml|table> # Output formatStart interactive REPL mode with boxed input interface.
$ wilk
┌──────────────────────────────────────┐
│ │
│ ██╗ ██╗██╗██╗ ██╗ ██╗ │
│ ██║ ██║██║██║ ██║ ██╔╝ │
│ ██║ █╗ ██║██║██║ █████╔╝ │
│ ██║███╗██║██║██║ ██╔═██╗ │
│ ╚███╔███╔╝██║███████╗██║ ██╗ │
│ ╚══╝╚══╝ ╚═╝╚══════╝╚═╝ ╚═╝ │
│ │
│ CLI-Native Agent Operating System │
│ │
└──────────────────────────────────────┘
Available agents: @research, @code-analyzer, @docs-writer
Memory loaded: 3 entries (847 tokens)
Type your message, or use @agent-name for specific agents
Press ↑/↓ for history, Tab for completion, Ctrl+C to exit
Tips for getting started:
1. Ask questions, edit files, or run commands
2. Be specific for the best results
3. Create WILK.md files to customize interactions
4. /help for more information
┌────────────────────────────────────────────────────────────────────────────┐
│ > Hello, help me understand this codebase │
└────────────────────────────────────────────────────────────────────────────┘
Send message to default agent or start conversation.
# Direct message
$ wilk chat "What's the weather like?"
# Interactive conversation
$ wilk chat
> Hello, how can I help you today?Create a new agent with interactive configuration.
$ wilk agent create research-assistant
Creating new agent 'research-assistant'...
Provider: [openai] anthropic, google, local
Model: [gpt-4] gpt-3.5-turbo, claude-3-sonnet, gemini-pro
Instructions: Research specialist focused on academic and technical topics
Tools: [web_search,file_ops] code_interpreter, rag_search
Memory: [enabled] disabled
Thinking display: [progressive] live, off
✅ Agent 'research-assistant' created (agent_abc123)Options:
--provider <name> # AI provider (openai, anthropic, google, local)
--model <name> # Model name
--instructions <text> # System instructions
--tools <list> # Comma-separated tool list
--no-memory # Disable memory for this agent
--thinking <mode> # Thinking display mode (off, progressive, live)
--collaborative # Allow others to edit
--template <name> # Use agent template
--from-file <path> # Create from YAML/JSON fileList all available agents.
$ wilk agent list
┌─────────────────┬──────────────┬─────────────┬──────────────┬─────────────┐
│ Agent │ Provider │ Model │ Tools │ Status │
├─────────────────┼──────────────┼─────────────┼──────────────┼─────────────┤
│ @research │ openai │ gpt-4 │ web, files │ Ready │
│ @code-analyzer │ anthropic │ claude-3.5 │ code, files │ Ready │
│ @docs-writer │ openai │ gpt-4 │ files, web │ Ready │
│ @security-audit │ anthropic │ claude-3 │ code, shell │ Installing │
└─────────────────┴──────────────┴─────────────┴──────────────┴─────────────┘
Total: 4 agents (3 ready, 1 installing)Options:
--filter <key=value> # Filter agents (provider=openai, status=ready)
--sort <field> # Sort by field (name, created, modified)
--format <table|json> # Output format
--verbose # Show detailed informationUpdate agent configuration with versioning.
$ wilk agent update research-assistant --instructions "Updated research specialist"
Updating agent 'research-assistant'...
Current version: 1.2.0
Changes detected:
- Instructions modified
- Tools unchanged
- Model parameters unchanged
Creating version 1.3.0...
✅ Agent updated to version 1.3.0Options:
--instructions <text> # Update system instructions
--model <name> # Change model
--tools <list> # Update tool list
--add-tools <list> # Add tools to existing list
--remove-tools <list> # Remove tools from list
--thinking <mode> # Update thinking display mode
--force-version # Force create new version even if no changes
--changelog <text> # Add changelog entryDelete an agent with confirmation.
$ wilk agent delete research-assistant
⚠️ This will permanently delete agent 'research-assistant'
Conversations: 15
Memory entries: 8
Projects using: 2
Continue? [y/N]: y
✅ Agent 'research-assistant' deletedView agent version history.
$ wilk agent versions research-assistant
Agent: research-assistant (agent_abc123)
┌─────────┬─────────────────────┬────────────┬─────────────────────────────┐
│ Version │ Created │ Author │ Changes │
├─────────┼─────────────────────┼────────────┼─────────────────────────────┤
│ 1.3.0 │ 2024-01-15 14:30:22 │ user123 │ Updated instructions │
│ 1.2.0 │ 2024-01-10 09:15:45 │ user123 │ Added web_search tool │
│ 1.1.0 │ 2024-01-08 16:20:10 │ user123 │ Model upgrade to gpt-4 │
│ 1.0.0 │ 2024-01-05 11:45:33 │ user123 │ Initial version │
└─────────┴─────────────────────┴────────────┴─────────────────────────────┘Revert agent to a previous version.
$ wilk agent revert research-assistant --version 1.2.0
Reverting agent 'research-assistant' to version 1.2.0...
Changes:
- Instructions: "Updated research specialist" → "Research specialist..."
- Tools: unchanged
- Model: unchanged
✅ Agent reverted to version 1.2.0 (new version 1.4.0 created)Execute specific agent with message.
$ @research-assistant "What are the latest developments in quantum computing?"
🤔 Agent thinking... (3.2s · ↑ 1.8k tokens)
● Task(Quantum Computing Research)
│ ⎿ Web Search("quantum computing 2024 developments")
│ Found 15 results (enter to expand)
│ ⎿ Analyzing latest papers...
│ ✓ 8 papers analyzed
│ ⎿ Synthesizing findings...
│ ✓ Research complete
🔬 Latest Quantum Computing Developments (2024)
Key breakthroughs this year include:
1. **IBM's 1000-qubit processor** - Achieved major milestone in December
2. **Google's quantum error correction** - Reduced error rates by 50%
3. **Microsoft's topological qubits** - New approach showing promise
[Detailed analysis continues...]
💭 Show agent reasoning? [y/N]:$ @research-assistant @security-audit "Analyze this new authentication protocol"
● Task(Multi-Agent Analysis)
│
├─ Agent(research-assistant)
│ ● Task(Protocol Research)
│ │ ⎿ Literature review...
│ │ Found 12 related papers (enter to expand)
│ │
├─ Agent(security-audit)
│ ● Task(Security Analysis)
│ │ ⎿ Threat modeling...
│ │ Identified 3 attack vectors (enter to expand)
│ │
● Task(Synthesis & Recommendations)
│ ⎿ Correlating findings...
│ ✓ Analysis complete
📋 Combined Analysis Results:
[Multi-agent synthesis...]Memory system management.
$ wilk memory
Current memory status:
Project Memory: ./WILK.md (847 tokens)
User Memory: ~/.wilk/memory/user.md (234 tokens)
Agent Memory: 3 agents with individual memory files
Quick actions:
wilk memory list - Show all memory entries
wilk memory edit - Edit project memory
wilk memory search - Search across memory
#<content> - Quick memory additionList all memory entries.
$ wilk memory list
Memory Overview:
📁 Project Memory (./WILK.md)
- Project overview and goals
- Architecture decisions
- Team members and roles
- Recent discussions (5 entries)
👤 User Memory (~/.wilk/memory/user.md)
- Personal preferences
- Frequently used commands
- Custom aliases
🤖 Agent Memory:
- @research-assistant: 8 learning entries
- @code-analyzer: 12 behavioral patterns
- @docs-writer: 3 style preferences
📄 Imported Files:
- @README.md (last updated: 2 hours ago)
- @package.json (last updated: 1 day ago)
- @src/types.ts (last updated: 3 hours ago)Search across all memory.
$ wilk memory search "authentication"
🔍 Searching memory for "authentication"...
Results (3 found):
📁 Project Memory (./WILK.md):
Line 23: "Authentication system uses JWT tokens with..."
Line 45: "OAuth integration planned for Q2..."
🤖 Agent Memory (@security-audit):
Entry 2: "Authentication patterns observed in codebase analysis..."
📄 Imported (@src/auth.ts):
Function: authenticateUser() - JWT validation logicEdit memory files with preferred editor.
$ wilk memory edit project
Opening ./WILK.md in VS Code...
$ wilk memory edit user
Opening ~/.wilk/memory/user.md in VS Code...
$ wilk memory edit agent @research-assistant
Opening ~/.wilk/agents/agent_abc123/memory.md in VS Code...Shortcut for adding content to memory.
$ #Remember: Use camelCase for all new TypeScript interfaces
Where should this be saved?
1. Project memory (./WILK.md)
2. User memory (~/.wilk/memory/user.md)
3. New file...
Choice [1]: 1
✅ Added to project memory under "Development Guidelines"List and manage conversation sessions.
$ wilk sessions
Recent Sessions:
┌──────────────┬─────────────────────┬───────────┬──────────────┬─────────────┐
│ Session ID │ Started │ Messages │ Agents │ Status │
├──────────────┼─────────────────────┼───────────┼──────────────┼─────────────┤
│ sess_abc123 │ 2024-01-15 14:30 │ 23 │ research │ Active │
│ sess_def456 │ 2024-01-15 12:15 │ 45 │ code,docs │ Paused │
│ sess_ghi789 │ 2024-01-14 16:20 │ 12 │ security │ Completed │
└──────────────┴─────────────────────┴───────────┴──────────────┴─────────────┘
Current session: sess_abc123 (23 messages, 1 agent)Resume a previous session with full context.
$ wilk session resume sess_def456
Resuming session sess_def456...
📋 Loading context:
- 45 messages (12.3k tokens)
- 2 agents: @code-analyzer, @docs-writer
- Memory context: 5 entries
- Tool executions: 8 completed
Last message (2024-01-15 12:45):
User: "Can you review the authentication module?"
@code-analyzer: "I've identified several areas for improvement..."
✅ Session resumed - continuing conversationSave current session with optional name.
$ wilk session save "auth-module-review"
Saving current session...
Session saved:
ID: sess_abc123
Name: "auth-module-review"
Messages: 23
Agents: @research-assistant
Memory: 3 entries
Duration: 2h 15m
💾 Session saved and can be resumed with:
wilk session resume sess_abc123Export session to file.
$ wilk session export sess_abc123 --format json
Exporting session sess_abc123...
Exported to: wilk-session-abc123.json
Messages: 23
Agent configs: 1
Memory context: included
Thinking content: included
File size: 45.2 KB
🔄 Import with: wilk session import wilk-session-abc123.jsonThinking system management.
$ wilk thinking
Agent Thinking Configuration:
Global Settings:
show_by_default: false
live_display: false
progressive_disclosure: true
export_thinking: true
Per-Agent Settings:
@research-assistant: progressive disclosure
@code-analyzer: live display
@docs-writer: show by default
Recent Thinking Sessions:
- sess_abc123: 3.2k thinking tokens, 5 reasoning chains
- sess_def456: 1.8k thinking tokens, 8 reasoning chainsDisplay thinking content from session.
$ wilk thinking show last
Last Agent Thinking (sess_abc123, @research-assistant):
┌─ Agent Reasoning ───────────────────────────────────────────────────────────────┐
│ I need to research quantum computing developments by: │
│ │
│ 1. Searching for recent papers and announcements │
│ - Focus on 2024 breakthroughs │
│ - Look for major tech companies (IBM, Google, Microsoft) │
│ │
│ 2. Analyzing the technical significance │
│ - Quantum error correction improvements │
│ - Qubit count milestones │
│ - Practical applications │
│ │
│ 3. Synthesizing findings into accessible explanation │
│ - Explain technical concepts clearly │
│ - Highlight practical implications │
│ - Provide future outlook │
└─────────────────────────────────────────────────────────────────────────────────┘
Thinking tokens: 1,847
Reasoning duration: 3.2 seconds
Model: gpt-4Export thinking content to file.
$ wilk thinking export sess_abc123
Exporting thinking content from session sess_abc123...
Exported:
- Agent reasoning chains: 5
- Total thinking tokens: 3,247
- File: thinking-export-sess_abc123.md
- Format: Markdown with metadata
📄 Thinking content saved to: thinking-export-sess_abc123.mdReplay session with thinking display.
$ wilk thinking replay sess_abc123
Replaying session sess_abc123 with thinking...
Message 1: "What are the latest developments in quantum computing?"
🤔 Agent thinking... (replaying 3.2s)
[Shows original thinking content with timing]
● Task(Quantum Computing Research)
[Replays entire task execution with thinking]Show current configuration.
$ wilk config
Wilk Configuration:
Global Settings:
default_provider: openai
default_model: gpt-4
max_context_tokens: 4096
context_strategy: discard
Agent Settings:
thinking_display: progressive
memory_integration: true
auto_save_sessions: true
CLI Settings:
editor: code
pager: less
output_format: table
colored_output: true
Paths:
config_dir: ~/.wilk
memory_dir: ~/.wilk/memory
agents_dir: ~/.wilk/agents
sessions_dir: ~/.wilk/sessionsSet configuration value.
$ wilk config set thinking.show_by_default true
✅ Set thinking.show_by_default = true
$ wilk config set default_model claude-3-sonnet
✅ Set default_model = claude-3-sonnet
$ wilk config set agent.@research.thinking live
✅ Set agent thinking mode for @research = liveGet specific configuration value.
$ wilk config get thinking.show_by_default
thinking.show_by_default = false
$ wilk config get agent.@research.model
agent.@research.model = gpt-4List available tools.
$ wilk tools
Available Tools:
Built-in Tools:
✅ file_operations - Read, write, and manage files
✅ shell_executor - Execute shell commands safely
✅ web_search - Search the web for information
✅ rag_search - Search project documentation
✅ code_interpreter - Execute Python/JavaScript code
MCP Tools:
✅ filesystem - Enhanced file operations
✅ github - GitHub repository integration
⚠️ database - Database query tool (needs config)
❌ slack - Slack integration (not installed)
Total: 8 tools (6 ready, 1 needs config, 1 not installed)MCP server management.
$ wilk mcp
Model Context Protocol (MCP) Status:
Active Servers:
✅ filesystem - Local file operations
✅ github - GitHub API integration
⚠️ database - PostgreSQL connector (config needed)
Available Servers:
📦 slack - Slack workspace integration
📦 notion - Notion API connector
📦 calendar - Google Calendar integration
Install with: wilk mcp add <server-name>Install MCP server.
$ wilk mcp add slack
Installing MCP server: slack...
📦 Installing @modelcontextprotocol/server-slack
🔧 Configuring server...
Configuration needed:
SLACK_BOT_TOKEN: Bot token for workspace access
SLACK_SIGNING_SECRET: Signing secret for verification
Set with: wilk mcp env slack SLACK_BOT_TOKEN your_token_here
✅ MCP server 'slack' installed (configuration required)Install agents, tools, or MCP servers.
$ wilk install @community/typescript-expert
Installing agent package: @community/typescript-expert...
📦 Downloading from registry...
🔍 Validating package integrity...
📋 Checking dependencies:
- Tools: code_interpreter, file_operations ✅
- Models: Compatible with gpt-4, claude-3 ✅
Installing:
Agent: typescript-expert
Templates: 3 conversation starters
Memory: TypeScript best practices
Tools: Custom TypeScript analysis tools
✅ Agent @typescript-expert installed and readySearch for agents, tools, or packages.
$ wilk search typescript
🔍 Searching for "typescript"...
Agents (3 found):
@community/typescript-expert - Advanced TypeScript development
@tools/ts-analyzer - TypeScript code analysis
@edu/typescript-tutor - TypeScript learning assistant
Tools (2 found):
typescript-compiler - Compile and validate TypeScript
ts-ast-analyzer - Abstract syntax tree analysis
MCP Servers (1 found):
typescript-language-server - TypeScript language server integrationSystem status and health check.
$ wilk status
Wilk System Status:
🟢 Core System:
Database: SQLite (healthy, 45.2 MB)
Memory: 89.3 MB used, 3.2 GB available
Agents: 4 loaded, 4 ready
Sessions: 1 active, 15 total
🟢 Providers:
OpenAI: Connected (quota: 85% remaining)
Anthropic: Connected (quota: 92% remaining)
Google: Not configured
Local: Ollama running (3 models loaded)
🟢 Tools:
Built-in: 5/5 ready
MCP: 3/4 ready (1 needs config)
🟡 Warnings:
- Database backup overdue (last: 3 days ago)
- Memory cleanup recommended
Overall: Healthy ✅Update Wilk and components.
$ wilk update
Checking for updates...
Updates available:
🔄 Wilk CLI: 1.2.3 → 1.3.0
🔄 Agent @research: 2.1.0 → 2.2.0
🔄 MCP server github: 1.0.5 → 1.1.0
Download size: 15.2 MB
Update all? [Y/n]: y
Updating...
✅ Wilk CLI updated to 1.3.0
✅ Agent @research updated to 2.2.0
✅ MCP server github updated to 1.1.0
🎉 All updates complete! Restart recommended.Diagnose and fix common issues.
$ wilk doctor
🩺 Running diagnostics...
✅ Configuration files: Valid
✅ Database integrity: OK
✅ Agent definitions: Valid (4 agents)
✅ Memory files: Accessible
✅ Provider connections: Online
⚠️ Tool permissions: Some tools need elevated access
❌ MCP server 'database': Configuration incomplete
Issues found: 2
🔧 Auto-fix available issues? [Y/n]: y
Fixing issues...
✅ Updated tool permissions
⚠️ MCP server 'database' requires manual configuration
Run: wilk mcp config database
Diagnostics complete. 1 manual action required.# Create agent and immediately test
$ wilk agent create test-agent --model gpt-3.5-turbo && @test-agent "Hello"
# Export session and thinking together
$ wilk session export sess_abc123 --with-thinking && wilk thinking export sess_abc123
# Batch update multiple agents
$ wilk agent list --filter provider=openai | xargs -I {} wilk agent update {} --model gpt-4# Search and install multiple packages
$ wilk search security | grep agent | head -3 | xargs -I {} wilk install {}
# Export multiple sessions
$ wilk sessions --status completed | jq -r '.id' | xargs -I {} wilk session export {}
# Find agents using specific tools
$ wilk agent list --format json | jq '.[] | select(.tools[] | contains("code_interpreter"))'# Switch to development profile
$ wilk --profile dev config set thinking.show_by_default true
# Production profile with restricted thinking
$ wilk --profile prod config set thinking.show_by_default false
$ wilk --profile prod config set agent.*.thinking offThis comprehensive command reference provides detailed documentation for all Wilk CLI commands, following LibreChat patterns while adding CLI-specific optimizations and the new thinking transparency features.