RomanNum3ral
/
cai
mirror of https://github.com/aliasrobotics/cai.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

docs: align REPL reference with current commands and navigation (#435) 

Remove council page and nav; drop obsolete slash commands from docs (/config,
/context, /platform, /kill, /model-show); document /env, /model show, and
CLI examples. Update README, MkDocs, and PRO pages for usage visibility.
This commit is contained in:
Yahya Hajji 2026-05-18 13:59:20 +02:00 committed by GitHub
parent 7ec0b4ccf2
commit 87cc46f46f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
21 changed files with 979 additions and 2191 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
README.md
View File

@ -1118,41 +1118,21 @@ Use ```/agent``` to list all the agents available.
<details> <details>
<summary> Where can I list all the environment variables? /config </summary> <summary> Where can I list all the environment variables? /env </summary>
![cai-008-config](imgs/readme_imgs/cai-008-config.png) ![cai-008-config](imgs/readme_imgs/cai-008-config.png)
</details> </details>
<details> <details>
<summary> How can I monitor context usage and token consumption? /context or /ctx 🚀 CAI PRO </summary> <summary> How can I monitor token usage and costs? </summary>
> **⚡ CAI PRO Exclusive Feature** Use **`/cost`** in the REPL for session spend and token statistics, **`/compact`** when conversations grow long, and (in **TUI** mode) the per-terminal cost and model indicators in the UI.
> The `/context` command is available exclusively in **[CAI PRO](https://aliasrobotics.com/cybersecurityai.php)**. To access this feature and unlock advanced monitoring capabilities, visit [Alias Robotics](https://aliasrobotics.com/cybersecurityai.php) for more information.
Use ```/context``` (or the short form ```/ctx```) to view your current context window usage and token statistics.
This command displays:
- Total context usage (used tokens / max tokens) with percentage
- Visual grid representation with the CAI logo showing filled context
- Detailed breakdown by category:
- System prompt tokens
- Tool definitions tokens
- Memory/RAG tokens
- User prompts tokens
- Assistant responses tokens
- Tool calls tokens
- Tool results tokens
- Free space available
**Why this matters**: Different models have different context limits (e.g., GPT-4: 128k tokens, Claude: 200k tokens). Monitoring your context usage helps you avoid hitting these limits during long conversations, which could cause errors or require conversation truncation.
```bash ```bash
# Show context usage CAI> /cost
/context
# Or use the short form
/ctx
``` ```
See the [CLI commands reference](docs/cli/commands_reference.md) for the full command list.
</details> </details>
<details> <details>

4
docs/api.md
View File

@ -10,7 +10,7 @@ cai --api --api-host 0.0.0.0 --api-port 8080
# the next free port and prints it in the console. # the next free port and prints it in the console.
``` ```
CLI flags and environment variables: CLI flags and environment variables (API subset):
| Flag | Env | Description | | Flag | Env | Description |
| --- | --- | --- | | --- | --- | --- |
@ -20,6 +20,8 @@ CLI flags and environment variables:
| `--api-reload` | `CAI_API_RELOAD` | Dev autoreload. | | `--api-reload` | `CAI_API_RELOAD` | Dev autoreload. |
| `--api-workers` | `CAI_API_WORKERS` | Worker processes (ignored with reload). | | `--api-workers` | `CAI_API_WORKERS` | Worker processes (ignored with reload). |
For **all** `cai` binary flags (`--tui`, `--resume`, `--yaml`, `--version`, …), see the single source of truth: [CLI commands reference — Binary `cai` CLI flags](cli/commands_reference.md#binary-cai-cli-flags).
Interactive docs at `/api/docs` and OpenAPI spec at `/api/openapi.json`. Interactive docs at `/api/docs` and OpenAPI spec at `/api/openapi.json`.
### Authentication ### Authentication

449
docs/cai/getting-started/commands.md
View File

@ -1,446 +1,13 @@
# CAI REPL Commands # CAI commands (hub)
This document provides documentation for all commands available in the CAI (Context-Aware Interface) REPL system. The **canonical** command reference (binary flags and REPL slash commands) lives here:
## Base Command System (`base.py`) - **[CLI commands reference](../../cli/commands_reference.md)** — use this for day-to-day lookup and GitHub Pages.
TUI-specific routing and shortcuts:
- **[TUI commands reference](../../tui/commands_reference.md)**
--- ---
## Core Commands *This hub replaces the long per-file command list to avoid drift; the CLI reference is updated with each framework release.*
### **Agent Management (`agent.py`)**
### **AgentCommand**
- **Command**: `/agent`
- **Purpose**: Managing and switching between different AI agents
- **Features**:
- List available agents
- Switch between agents
- Display agent information
- Visualize agent interaction graphs
### **Configuration Management (`config.py`)**
### **ConfigCommand**
- **Command**: `/config`
- **Purpose**: Display and configure environment variables
- **Features**:
- Show current environment variable settings
- Configure CTF (Capture The Flag) variables
- Manage configuration through environment variables
### **Context Usage Monitoring (`context.py`)** 🚀 **CAI PRO Exclusive**
### **ContextCommand**
> **⚡ CAI PRO Exclusive Feature**
> The `/context` command is available exclusively in **[CAI PRO](https://aliasrobotics.com/cybersecurityai.php)**. To access this feature and unlock advanced monitoring capabilities, visit [Alias Robotics](https://aliasrobotics.com/cybersecurityai.php) for more information.
- **Command**: `/context` or `/ctx`
- **Purpose**: View context usage and token statistics for the current conversation
- **Features**:
- Display total context usage (used/max tokens) with percentage
- Visual grid representation of context usage with CAI logo
- Detailed breakdown by category:
- System prompt tokens
- Tool definitions tokens
- Memory/RAG tokens
- User prompts tokens
- Assistant responses tokens
- Tool calls tokens
- Tool results tokens
- Free space visualization
- Context usage tracking across conversation history
- Real-time token consumption monitoring
- **Usage Example**:
```bash
# Show context usage for current agent
/context
# Alternative short form
/ctx
```
- **Output Includes**:
- Visual grid showing filled vs free context space
- Percentage breakdown per category
- Token counts formatted with 'k' suffix for thousands
- Color-coded categories for easy identification
- Summary of total input tokens from last iteration
### **Cost Tracking (`cost.py`)**
### **CostCommand**
- **Command**: `/cost`
- **Purpose**: View usage costs and statistics
- **Features**:
- Display current session costs
- Show cost breakdowns by model
- Track usage over time
- Cost statistics and reporting
### **Exit (`exit.py`)**
### **ExitCommand**
- **Command**: `/exit`
- **Purpose**: Terminate the CAI REPL session
- **Features**:
- Clean shutdown of the REPL
- Save current session data
- Cleanup background processes
### **Help System (`help.py`)**
### **HelpCommand**
- **Command**: `/help` or `/?`
- **Purpose**: Display help information and command documentation
- **Features**:
- List available commands
- Show command usage
- Display command aliases
- Provide help for specific commands
### **History Management (`history.py`)**
### **HistoryCommand**
- **Command**: `/history`
- **Purpose**: Display conversation history with agent filtering
- **Features**:
- Show conversation history
- Filter by specific agents
- Display message tree structure
- Export history functionality
---
## Data Management Commands
### **Compact Conversation (`compact.py`)**
### **CompactCommand**
- **Command**: `/compact`
- **Purpose**: Compact current conversation and manage model/prompt settings
- **Features**:
- Reduce conversation context size
- Change model during compaction
- Modify prompt settings
- Maintain conversation flow while reducing tokens
### **Environment Display (`env.py`)**
### **EnvCommand**
- **Command**: `/env`
- **Purpose**: Display current environment variables
- **Features**:
- Show all environment variables
- Filter by variable patterns
- Display CAI-specific environment settings
### **Load Data (`load.py`)**
### **LoadCommand**
- **Command**: `/load`
- **Purpose**: Load JSONL data into the current session context
- **Features**:
- Load conversation history from files
- Import external data
- Integrate with parallel configurations
- Support for various data formats
### **Memory Management (`memory.py`)**
### **MemoryCommand**
- **Command**: `/memory`
- **Purpose**: Manage persistent memory storage in `.cai/memory`
- **Features**:
- Store conversation context persistently
- Apply memory to current context
- Manage memory entries
- Persistent storage across sessions
### **Flush History (`flush.py`)**
### **FlushCommand**
- **Command**: `/flush`
- **Purpose**: Clear conversation history
- **Features**:
- Clear current conversation
- Reset agent contexts
- Clean up memory
- Start fresh conversation
---
## Model Management Commands
### **Model Configuration (`model.py`)**
### **ModelCommand**
- **Command**: `/model`
- **Purpose**: View and change the current LLM model
- **Features**:
- Switch between different models
- Display model information
- Configure model parameters
- Support for LiteLLM and Ollama
### **ModelShowCommand**
- **Command**: `/model-show`
- **Purpose**: Show all available models from LiteLLM repository
- **Features**:
- List all available models
- Display model categories
- Show model capabilities
- Filter by provider
---
## Advanced Features
### **Graph Visualization (`graph.py`)**
### **GraphCommand**
- **Command**: `/graph`
- **Purpose**: Visualize agent interaction graphs
- **Features**:
- Display directed graph of conversations
- Show user and agent interactions
- Highlight tool calls
- Visualize conversation flow
### **Parallel Execution (`parallel.py`)**
### **ParallelCommand**
- **Command**: `/parallel`
- **Purpose**: Configure and run parallel agent workflows with isolated contexts
- **Features**:
- Add/remove/list parallel agents
- Queue prompts per agent or broadcast to all agents
- Execute queued prompts with `/parallel run`
- Merge results back into the main context
- Exit parallel mode with or without merge
### **Queue Management (`queue.py`)**
### **QueueCommand**
- **Command**: `/queue`
- **Purpose**: Manage sequential prompt queue independently from parallel mode
- **Features**:
- Add prompts to queue
- List queued prompts
- Run queued prompts sequentially
- Clear queue safely
### **Merge Histories (`merge.py`)**
### **MergeCommand**
- **Command**: `/merge`
- **Purpose**: Merge parallel agent contexts into main context and exit parallel mode
- **Features**:
- Combine histories from multiple agents
- Integrate parallel conversation results into the current main thread
- Automatically leave parallel mode after successful merge
---
## Integration Commands
### **MCP Integration (`mcp.py`)**
### **MCPCommand**
- **Command**: `/mcp`
- **Purpose**: Manage MCP (Model Context Protocol) servers and their tools
- **Features**:
- Load SSE MCP servers
- Load STDIO MCP servers
- List active MCP connections
- Add MCP tools to agents
- Manage MCP server lifecycle
### **Platform Features (`platform.py`)**
### **PlatformCommand**
- **Command**: `/platform`
- **Purpose**: Interact with platform-specific features
- **Features**:
- Access platform extensions
- Platform-specific integrations
- Check platform availability
---
## System Management Commands
### **Process Management (`kill.py`)**
### **KillCommand**
- **Command**: `/kill`
- **Purpose**: Terminate active processes or sessions
- **Features**:
- Kill background processes
- Terminate stuck sessions
- Process cleanup
### **Shell Access (`shell.py`)**
### **ShellCommand**
- **Command**: `/shell`
- **Purpose**: Execute shell commands from within the REPL
- **Features**:
- Run system commands
- Access workspace directory
- Container workspace support
- Signal handling for processes
### **Virtualization (`virtualization.py`)**
### **VirtualizationCommand**
- **Command**: `/virtualization` or `/virt`
- **Purpose**: Manage Docker-based virtualization environments
- **Features**:
- Set up Docker containers
- Manage container lifecycle
- Workspace virtualization
- Environment isolation
### **Workspace Management (`workspace.py`)**
### **WorkspaceCommand**
- **Command**: `/workspace` or `/ws`
- **Purpose**: Manage workspace within Docker containers or locally
- **Features**:
- Navigate workspace directories
- Mount external directories
- Container workspace management
- File system operations
### **Quickstart (`quickstart.py`)**
### **QuickstartCommand**
- **Command**: `/quickstart`
- **Purpose**: Display setup information for new users
- **Features**:
- Essential setup guidance
- Configuration instructions
- Getting started tutorial
- Auto-runs on first launch
---
## Utility Commands
### **Command Completion (`completer.py`)**
### **FuzzyCommandCompleter**
- **Purpose**: Intelligent command completion with fuzzy matching
- **Features**:
- Command auto-completion
- Fuzzy matching for typos
- Subcommand suggestions
- Argument completion
- Command shadowing detection
---
## Usage Examples
### Basic Workflow
```bash
# Start CAI REPL
cai
# View available agents
/agent list
# Switch to a specific agent
/agent switch <agent_name>
# View conversation history
/history
# Change model
/model gpt-4
# Clear conversation
/flush
# Exit
/exit
```
### Advanced Features
```bash
# Set up parallel execution
/parallel add red_teamer
/parallel add network_traffic_analyzer
# Add prompts (per agent or all)
/parallel prompt all "Scan 192.168.1.0/24"
# Execute in parallel
/parallel run
# Merge all parallel contexts into main context and exit parallel mode
/merge
# Optional: exit without merging contexts
/parallel clear
```
### Integration Examples
```bash
# Load MCP server
/mcp load http://localhost:9876/sse burp
# Add MCP tools to agent
/mcp add-to-agent <agent_name> burp
# Set up virtualized environment
/virtualization create ubuntu:latest
/workspace /path/to/project
```
---
## Command Registration
All commands are automatically registered when their respective modules are imported through the `__init__.py` file. The command system uses a registry pattern to track all available commands and their aliases.
---
## File Structure
```
src/cai/repl/commands/
├── __init__.py # Module exports and imports
├── base.py # Base command class
├── agent.py # Agent management
├── compact.py # Conversation compaction
├── completer.py # Command completion
├── config.py # Configuration management
├── cost.py # Cost tracking
├── env.py # Environment variables
├── exit.py # REPL exit
├── flush.py # History clearing
├── graph.py # Graph visualization
├── help.py # Help system
├── history.py # History management
├── kill.py # Process management
├── load.py # Data loading
├── mcp.py # MCP integration
├── memory.py # Memory management
├── merge.py # History merging
├── model.py # Model management
├── parallel.py # Parallel execution
├── platform.py # Platform features
├── quickstart.py # User onboarding
├── run.py # Parallel execution trigger
├── shell.py # Shell access
├── virtualization.py # Container management
└── workspace.py # Workspace management
```
---
## Extending the Command System
To add new commands:
1. Create a new Python file in `src/cai/repl/commands/`
2. Import the base `Command` class from `base.py`
3. Extend the `Command` class with your implementation
4. Use the `register_command` decorator or function
5. Add the import to `__init__.py`
Example:
```python
from cai.repl.commands.base import Command, register_command
class MyCommand(Command):
def __init__(self):
super().__init__(
name="/mycommand",
description="My custom command",
aliases=["/my", "/mc"]
)
def execute(self, args):
# Command implementation
pass
register_command(MyCommand())
```

2
docs/cai/getting-started/configuration.md
View File

@ -16,7 +16,7 @@ The OPENAI_API_KEY must not be left blank. It should contain either "sk-123" (as
For a complete reference organized by use case, see [Environment Variables Reference](../../environment_variables.md). For a complete reference organized by use case, see [Environment Variables Reference](../../environment_variables.md).
**In the REPL:** `/config` (or `/config list`) shows variables with **current values** and index numbers. **`/help`** includes the **full environment reference** (tables) after the quick guide; **`/help var NAME`** opens **long-form** help for a single variable. See also [Environment Variables — Discovering variables in the REPL](../../environment_variables.md#discovering-variables-in-the-repl). **In the REPL:** `/env list` shows the catalog with **current values** and index numbers; bare `/env` shows `CAI_*` / `CTF_*` in this session. **`/help`** includes the **full environment reference** (tables) after the quick guide; **`/help var NAME`** opens **long-form** help for a single variable. See also [Environment Variables — Discovering variables in the REPL](../../environment_variables.md#discovering-variables-in-the-repl).
| Variable | Description | Default | | Variable | Description | Default |
|----------|-------------|---------| |----------|-------------|---------|

4
docs/cai_faq.md
View File

@ -84,9 +84,9 @@
![cai-010-agents-menu](media/cai-010-agents-menu.png) ![cai-010-agents-menu](media/cai-010-agents-menu.png)
??? question "Where can I list all the environment variables? /config" ??? question "Where can I list all the environment variables? /env"
Use **`/config`** (or **`/config list`**) to see all variables with **current values** and index numbers for `/config set`. Use **`/env list`** to see the full catalog with **current values** and index numbers for **`/env set`**. Bare **`/env`** shows only `CAI_*` / `CTF_*` keys in the current session.
For **full documentation tables** (defaults, constraints, when each applies), run **`/help`** and scroll past the quick guide, or **`/help topics`** and read to the end. For **one variable** in depth, use **`/help var VARIABLE_NAME`** (e.g. `/help var CAI_DEBUG`). For **full documentation tables** (defaults, constraints, when each applies), run **`/help`** and scroll past the quick guide, or **`/help topics`** and read to the end. For **one variable** in depth, use **`/help var VARIABLE_NAME`** (e.g. `/help var CAI_DEBUG`).

113
docs/cai_pro.md
View File

@ -19,39 +19,41 @@ The cybersecurity AI landscape is rapidly evolving, and professionals need tools
- **🇪🇺 European Hosting**: GDPR and NIS2 compliant infrastructure ensuring maximum privacy and data sovereignty - **🇪🇺 European Hosting**: GDPR and NIS2 compliant infrastructure ensuring maximum privacy and data sovereignty
- **💬 Professional Support**: Dedicated technical support to help you maximize your security testing capabilities - **💬 Professional Support**: Dedicated technical support to help you maximize your security testing capabilities
- **📱 Mobile UI (iOS)**: Native iOS app for security testing on the go - **[Join TestFlight Beta](https://testflight.apple.com/join/nXZZD4Z5)** - **📱 Mobile UI (iOS)**: Native iOS app for security testing on the go - **[Join TestFlight Beta](https://testflight.apple.com/join/nXZZD4Z5)**
- **⚡ Advanced Features**: Terminal UI (deprecated), context monitoring, and exclusive tools not available in the Community Edition - **⚡ Advanced Features**: Terminal UI (CLI recommended), usage and cost tooling (`/cost`, compaction), and exclusive tools not available in the Community Edition
--- ---
## CAI FREE vs CAI PRO ## CAI FREE vs CAI PRO
<div class="comparison-table" markdown>
| Feature | CAI FREE | CAI PRO |
|---------|-------------------|---------|
| **Core Framework** | ✅ Access, ~6-months behind | ✅ Full Access |
| **300+ AI Models** | ✅ BYO API Keys | ✅ BYO API Keys |
| **Built-in Security Tools** | ✅ Full Suite | ✅ Full Suite |
| **Agent Patterns** | ✅ All Patterns | ✅ All Patterns |
| **Command Line Interface** | ✅ Yes | ✅ Yes |
| **European Data Hosting** | ✅ **GDPR + NIS2 Compliant** | ✅ **GDPR + NIS2 Compliant** |
| **`alias1` Model** | ❌ Not Available | ✅ **Unlimited Tokens** |
| **Mobile UI (iOS App)** | ❌ | ✅ **Native iOS App** - [TestFlight](https://testflight.apple.com/join/nXZZD4Z5) |
| **Terminal User Interface (TUI)** | ❌ | ✅ Multi-terminal (Deprecated) |
| **Context Monitoring (`/context`)** | ❌ | ✅ Real-time token tracking |
| **Advanced Reporting** | ❌ | ✅ Professional formats |
| **Priority Support** | ❌ Community | ✅ **Professional** |
| **Commercial Use License** | ❌ Research Only | ✅ **Full Commercial** |
| **Custom Extensions** | ❌ | ✅ Available on request |
| **Pricing** | **Free** (Research) | **€350/month** |
</div>
| Feature | CAI FREE | CAI PRO |
| ----------------------------------- | --------------------------- | ------------------------------------------------------------------------------- |
| **Core Framework** | ✅ Access, ~6-months behind | ✅ Full Access |
| **300+ AI Models** | ✅ BYO API Keys | ✅ BYO API Keys |
| **Built-in Security Tools** | ✅ Full Suite | ✅ Full Suite |
| **Agent Patterns** | ✅ All Patterns | ✅ All Patterns |
| **Command Line Interface** | ✅ Yes | ✅ Yes |
| **European Data Hosting** | ✅ **GDPR + NIS2 Compliant** | ✅ **GDPR + NIS2 Compliant** |
| `**alias1` Model** | ❌ Not Available | ✅ **Unlimited Tokens** |
| **Mobile UI (iOS App)** | ❌ | ✅ **Native iOS App** - [TestFlight](https://testflight.apple.com/join/nXZZD4Z5) |
| **Terminal User Interface (TUI)** | ❌ | ✅ Multi-terminal (Deprecated) |
| **Usage & cost visibility** | ❌ | ✅ `/cost`, TUI cost panels, reporting workflows |
| **Advanced Reporting** | ❌ | ✅ Professional formats |
| **Priority Support** | ❌ Community | ✅ **Professional** |
| **Commercial Use License** | ❌ Research Only | ✅ **Full Commercial** |
| **Custom Extensions** | ❌ | ✅ Available on request |
| **Pricing** | **Free** (Research) | **€350/month** |
--- ---
## The `alias1` Model ## The `alias1` Model
<div class="highlight-box" markdown>
### 🏆 **Beats GPT-5 in Cybersecurity Benchmarks** ### 🏆 **Beats GPT-5 in Cybersecurity Benchmarks**
@ -63,11 +65,12 @@ The `alias1` model is our flagship cybersecurity AI, specifically trained for:
- **Zero Refusals**: No ethical restrictions for authorized security testing - **Zero Refusals**: No ethical restrictions for authorized security testing
**Performance Highlights:** **Performance Highlights:**
- Outperforms GPT-5 in AI vs AI cybersecurity benchmarks - Outperforms GPT-5 in AI vs AI cybersecurity benchmarks
- 500B-parameter architecture optimized for security workflows - 500B-parameter architecture optimized for security workflows
- Unrestricted responses for authorized pentesting engagements - Unrestricted responses for authorized pentesting engagements
</div>
[View Full Benchmarks →](https://aliasrobotics.com/alias1.php#benchmarking) [View Full Benchmarks →](https://aliasrobotics.com/alias1.php#benchmarking)
@ -75,20 +78,16 @@ The `alias1` model is our flagship cybersecurity AI, specifically trained for:
The performance of `alias1` and the CAI framework is validated through rigorous peer-reviewed research: The performance of `alias1` and the CAI framework is validated through rigorous peer-reviewed research:
- 📊 [**CAIBench: Cybersecurity AI Benchmark**](https://arxiv.org/pdf/2510.24317) (2025) - 📊 **[CAIBench: Cybersecurity AI Benchmark](https://arxiv.org/pdf/2510.24317)** (2025)
Modular meta-benchmark framework for evaluating LLM models across offensive and defensive cybersecurity domains. `alias1` demonstrates superior performance compared to general-purpose models. Modular meta-benchmark framework for evaluating LLM models across offensive and defensive cybersecurity domains. `alias1` demonstrates superior performance compared to general-purpose models.
- 🎯 **[Evaluating Agentic Cybersecurity in Attack/Defense CTFs](https://arxiv.org/pdf/2510.17521)** (2025)
- 🎯 [**Evaluating Agentic Cybersecurity in Attack/Defense CTFs**](https://arxiv.org/pdf/2510.17521) (2025) Real-world evaluation showing defensive agents achieved 54.3% patching success versus 28.3% offensive initial access. Validates practical effectiveness of CAI agents in live CTF environments.
Real-world evaluation showing defensive agents achieved 54.3% patching success versus 28.3% offensive initial access. Validates practical effectiveness of CAI agents in live CTF environments. - 🚀 **[Cybersecurity AI (CAI) Framework](https://arxiv.org/pdf/2504.06017)** (2025)
Core framework paper demonstrating that CAI outperforms humans by up to **3,600× in specific security testing scenarios**, establishing a new standard for automated security assessment.
- 🚀 [**Cybersecurity AI (CAI) Framework**](https://arxiv.org/pdf/2504.06017) (2025) - 🛡️ **[Hacking the AI Hackers via Prompt Injection](https://arxiv.org/pdf/2508.21669)** (2025)
Core framework paper demonstrating that CAI outperforms humans by up to **3,600× in specific security testing scenarios**, establishing a new standard for automated security assessment. Demonstrates four-layer guardrail defenses against prompt injection attacks, ensuring `alias1` remains secure even when processing adversarial inputs.
- 📚 **[CAI Fluency: Educational Framework](https://arxiv.org/pdf/2508.13588)** (2025)
- 🛡️ [**Hacking the AI Hackers via Prompt Injection**](https://arxiv.org/pdf/2508.21669) (2025) Comprehensive educational platform for democratizing cybersecurity AI knowledge and best practices.
Demonstrates four-layer guardrail defenses against prompt injection attacks, ensuring `alias1` remains secure even when processing adversarial inputs.
- 📚 [**CAI Fluency: Educational Framework**](https://arxiv.org/pdf/2508.13588) (2025)
Comprehensive educational platform for democratizing cybersecurity AI knowledge and best practices.
**Explore all research:** [Alias Robotics Research Library](https://aliasrobotics.com/research-security.php#papers) (24+ peer-reviewed publications) **Explore all research:** [Alias Robotics Research Library](https://aliasrobotics.com/research-security.php#papers) (24+ peer-reviewed publications)
@ -128,12 +127,14 @@ Building and maintaining CAI is resource-intensive. CAI PRO enables us to:
### Fair Pricing ### Fair Pricing
**€350/month** provides: **€350/month** provides:
- **Unlimited `alias1` tokens** (compare: OpenAI GPT-4o costs ~$2.50 per 1M tokens) - **Unlimited `alias1` tokens** (compare: OpenAI GPT-4o costs ~$2.50 per 1M tokens)
- **Professional support** (compare: enterprise support typically $1000+/month) - **Professional support** (compare: enterprise support typically $1000+/month)
- **Privacy guarantees** (priceless for security professionals) - **Privacy guarantees** (priceless for security professionals)
- **Commercial license** (required for security consulting businesses) - **Commercial license** (required for security consulting businesses)
Most security professionals already pay similar or higher amounts for: Most security professionals already pay similar or higher amounts for:
- **Burp Suite Professional**: $449/year ($37/month) - **Burp Suite Professional**: $449/year ($37/month)
- **ChatGPT Plus/Pro**: $20-200/month (with severe restrictions) - **ChatGPT Plus/Pro**: $20-200/month (with severe restrictions)
- **Other AI security tools**: $500-2000/month (closed-source, inferior models) - **Other AI security tools**: $500-2000/month (closed-source, inferior models)
@ -155,16 +156,9 @@ Run multiple agents in parallel with an intuitive multi-terminal interface:
[TUI Documentation →](tui/tui_index.md) [TUI Documentation →](tui/tui_index.md)
### 📊 Context Monitoring (`/context`) ### 📊 Usage and long threads
Track token usage and optimize your conversations: Track spend with **`/cost`**, use **`/compact`** when conversations grow large, and use the **TUI** cost/model indicators for per-terminal visibility.
- **Real-time Tracking**: Monitor context window consumption as you work
- **Category Breakdown**: See tokens by system, tools, memory, and messages
- **Visual Grid**: CAI logo-based visualization of context utilization
- **Optimization Insights**: Understand when to compact or clear history
[Context Command Docs →](cai/getting-started/commands.md#context-usage-monitoring-contextpy-cai-pro-exclusive)
### 📝 Advanced Reporting ### 📝 Advanced Reporting
@ -205,11 +199,13 @@ CAI_GUARDRAILS=true
### 3. Launch CAI PRO ### 3. Launch CAI PRO
#### CLI Mode (Standard) #### CLI Mode (Standard)
```bash ```bash
cai cai
``` ```
#### TUI Mode (Multi-terminal) #### TUI Mode (Multi-terminal)
```bash ```bash
cai --tui cai --tui
``` ```
@ -222,10 +218,10 @@ Check that you're using CAI PRO features:
CAI> /model CAI> /model
# Should show alias1 is available # Should show alias1 is available
CAI> /context CAI> /cost
# Should display context usage # Should display session usage / costs
CAI> --tui cai --tui
# Should launch multi-terminal interface # Should launch multi-terminal interface
``` ```
@ -237,7 +233,7 @@ CAI> --tui
**CAI PRO subscribers receive:** **CAI PRO subscribers receive:**
- **Email Support**: research@aliasrobotics.com (48-hour response SLA) - **Email Support**: [research@aliasrobotics.com](mailto:research@aliasrobotics.com) (48-hour response SLA)
- **Priority Discord**: Exclusive #pro-support channel - **Priority Discord**: Exclusive #pro-support channel
- **Quarterly Strategy Calls**: Discuss roadmap and feature requests - **Quarterly Strategy Calls**: Discuss roadmap and feature requests
- **Custom Development**: Request tailored agents and extensions - **Custom Development**: Request tailored agents and extensions
@ -246,7 +242,7 @@ CAI> --tui
- **[TUI Guide](tui/tui_index.md)**: Complete Terminal UI documentation - **[TUI Guide](tui/tui_index.md)**: Complete Terminal UI documentation
- **[Agent Reference](agents.md)**: All available agents and configurations - **[Agent Reference](agents.md)**: All available agents and configurations
- **[Command Reference](cai/getting-started/commands.md)**: Full CLI/TUI command list - **[Command Reference](cli/commands_reference.md)**: Full CLI/TUI command list
- **[Benchmarks](cai_benchmark.md)**: Performance data and comparisons - **[Benchmarks](cai_benchmark.md)**: Performance data and comparisons
### Community Resources ### Community Resources
@ -276,7 +272,8 @@ No. The Community Edition license restricts use to research and educational purp
### Do you offer team/enterprise pricing? ### Do you offer team/enterprise pricing?
Yes! Contact research@aliasrobotics.com for: Yes! Contact [research@aliasrobotics.com](mailto:research@aliasrobotics.com) for:
- **Team plans** (5+ users): Volume discounts - **Team plans** (5+ users): Volume discounts
- **Enterprise plans** (20+ users): Custom pricing, on-premise deployment - **Enterprise plans** (20+ users): Custom pricing, on-premise deployment
- **Academic licenses**: Special rates for universities and research institutions - **Academic licenses**: Special rates for universities and research institutions
@ -284,6 +281,7 @@ Yes! Contact research@aliasrobotics.com for:
### Is my security testing data private? ### Is my security testing data private?
**Absolutely.** CAI PRO guarantees: **Absolutely.** CAI PRO guarantees:
- **No training on your data**: Your pentesting activities never improve our models (unless you explicitly opt in) - **No training on your data**: Your pentesting activities never improve our models (unless you explicitly opt in)
- **European hosting**: All data processed in GDPR-compliant datacenters - **European hosting**: All data processed in GDPR-compliant datacenters
- **No third-party sharing**: Unlike OpenAI/Anthropic, we never send your data elsewhere - **No third-party sharing**: Unlike OpenAI/Anthropic, we never send your data elsewhere
@ -292,6 +290,7 @@ Yes! Contact research@aliasrobotics.com for:
### Can I switch between models? ### Can I switch between models?
Yes! CAI PRO includes: Yes! CAI PRO includes:
- **Unlimited `alias1` tokens** (your PRO model) - **Unlimited `alias1` tokens** (your PRO model)
- **BYO API keys**: Continue using OpenAI, Anthropic, etc. with your own keys - **BYO API keys**: Continue using OpenAI, Anthropic, etc. with your own keys
- **Mix and match**: Use `alias1` for exploitation, GPT-4 for reporting, etc. - **Mix and match**: Use `alias1` for exploitation, GPT-4 for reporting, etc.
@ -299,6 +298,7 @@ Yes! CAI PRO includes:
### What if alias1 refuses a query? ### What if alias1 refuses a query?
`alias1` has **zero refusals** for authorized security testing. If you encounter issues: `alias1` has **zero refusals** for authorized security testing. If you encounter issues:
1. Ensure your prompt includes security context (e.g., "authorized pentest of...") 1. Ensure your prompt includes security context (e.g., "authorized pentest of...")
2. Check your `CAI_GUARDRAILS` setting (may block malicious patterns) 2. Check your `CAI_GUARDRAILS` setting (may block malicious patterns)
3. Contact support—we'll investigate immediately 3. Contact support—we'll investigate immediately
@ -309,13 +309,13 @@ Yes! CAI PRO includes:
**Transform your security testing workflow with CAI PRO.** **Transform your security testing workflow with CAI PRO.**
<div class="cta-box" markdown>
### 🚀 **Ready to Upgrade?** ### 🚀 **Ready to Upgrade?**
- ✅ Unlimited `alias1` access - ✅ Unlimited `alias1` access
- ✅ Terminal UI with parallel agents - ✅ Terminal UI with parallel agents
- ✅ Context monitoring and optimization - ✅ Usage and cost visibility (`/cost`, compaction, TUI)
- ✅ Professional support - ✅ Professional support
- ✅ European data privacy - ✅ European data privacy
- ✅ Commercial use license - ✅ Commercial use license
@ -324,11 +324,8 @@ Yes! CAI PRO includes:
**[Get CAI PRO →](https://aliasrobotics.com/cybersecurityai.php)** **[Get CAI PRO →](https://aliasrobotics.com/cybersecurityai.php)**
</div>
--- ---
<small> *Have questions? Contact research@aliasrobotics.com* *Need a quote for your organization? [Request enterprise pricing →](mailto:research@aliasrobotics.com?subject=CAI%20PRO%20Enterprise%20Inquiry)*
*Have questions? Contact research@aliasrobotics.com*
*Need a quote for your organization? [Request enterprise pricing →](mailto:research@aliasrobotics.com?subject=CAI%20PRO%20Enterprise%20Inquiry)*
</small>

44
docs/cai_pro_features.md
View File

@ -14,7 +14,7 @@
|------------|----------|---------| |------------|----------|---------|
| **🤖 Alias1 Model** | ❌ | ✅ **Unlimited Tokens** | | **🤖 Alias1 Model** | ❌ | ✅ **Unlimited Tokens** |
| **🖥️ Terminal UI (TUI)** | ❌ | ✅ Multi-terminal parallel execution | | **🖥️ Terminal UI (TUI)** | ❌ | ✅ Multi-terminal parallel execution |
| **📊 Context Monitoring** | ❌ | ✅ Real-time token tracking | | **📊 Usage / cost** | ❌ | ✅ `/cost`, TUI cost panels |
| **⚡ Multi-Agent Swarms** | ❌ | ✅ 100+ parallel agents | | **⚡ Multi-Agent Swarms** | ❌ | ✅ 100+ parallel agents |
| **💬 Professional Support** | ❌ Community | ✅ Priority (48h SLA) | | **💬 Professional Support** | ❌ Community | ✅ Priority (48h SLA) |
| **🇪🇺 European Hosting** | ✅ GDPR + NIS2 | ✅ GDPR + NIS2 | | **🇪🇺 European Hosting** | ✅ GDPR + NIS2 | ✅ GDPR + NIS2 |
@ -59,7 +59,7 @@
#### Keyboard Control #### Keyboard Control
- **Vim-style shortcuts**: Navigate without touching your mouse - **Vim-style shortcuts**: Navigate without touching your mouse
- **Quick commands**: `/agent`, `/model`, `/context`, `/parallel` - **Quick commands**: `/agent`, `/model`, `/parallel`, `/cost`
- **Terminal switching**: `Ctrl+N`/`Ctrl+B` for rapid navigation - **Terminal switching**: `Ctrl+N`/`Ctrl+B` for rapid navigation
#### Real-Time Stats #### Real-Time Stats
@ -83,29 +83,7 @@
--- ---
## 3. Context Monitoring (`/context`) ## 3. Parallel Agent Swarms
### Optimize Your Conversations
Track token usage and stay within model limits with real-time context monitoring.
### Features
- **Real-time tracking**: Monitor context window consumption as you work
- **Category breakdown**: Tokens by system, tools, memory, and messages
- **Visual indicators**: Color-coded utilization levels
- **Optimization insights**: Know when to compact or clear history
### Use Cases
- **Long conversations**: Avoid hitting context limits mid-session
- **Cost optimization**: Understand where tokens are consumed
- **Memory management**: Balance RAG memory vs conversation space
- **Multi-terminal coordination**: Track context across parallel agents
---
## 4. Parallel Agent Swarms
### 100+ Concurrent Agents ### 100+ Concurrent Agents
@ -148,7 +126,7 @@ Chain agents for complex operations:
--- ---
## 5. Professional Support ## 4. Professional Support
### Priority Technical Assistance ### Priority Technical Assistance
@ -181,7 +159,7 @@ Request tailored solutions:
--- ---
## 6. European Data Sovereignty ## 5. European Data Sovereignty
### GDPR & NIS2 Compliant by Design ### GDPR & NIS2 Compliant by Design
@ -207,7 +185,7 @@ Your security testing data never leaves Europe.
--- ---
## 7. Advanced Reporting ## 6. Advanced Reporting
### Professional Security Reports ### Professional Security Reports
@ -242,7 +220,7 @@ Generate compliance-ready reports automatically.
--- ---
## 8. Four-Layer Guardrails ## 7. Four-Layer Guardrails
### Advanced Security Protection ### Advanced Security Protection
@ -274,7 +252,7 @@ CAI PRO includes enterprise-grade guardrails against adversarial attacks.
--- ---
## 9. Commercial Use License ## 8. Commercial Use License
### Unrestricted Business Use ### Unrestricted Business Use
@ -302,7 +280,7 @@ CAI PRO includes full commercial licensing for professional security services.
--- ---
## 10. Custom Extensions ## 9. Custom Extensions
### Tailored Solutions for Your Organization ### Tailored Solutions for Your Organization
@ -344,7 +322,7 @@ Work with the CAI team to develop specialized capabilities.
| **300+ Models** | ✅ BYO Keys | ✅ BYO Keys | ✅ BYO Keys + Private | | **300+ Models** | ✅ BYO Keys | ✅ BYO Keys | ✅ BYO Keys + Private |
| **Alias1 Tokens** | ❌ | ✅ Unlimited | ✅ Unlimited + On-prem | | **Alias1 Tokens** | ❌ | ✅ Unlimited | ✅ Unlimited + On-prem |
| **TUI** | ❌ | ✅ Yes | ✅ Yes + Custom UI | | **TUI** | ❌ | ✅ Yes | ✅ Yes + Custom UI |
| **Context Monitoring** | ❌ | ✅ Yes | ✅ Yes + Analytics | | **Usage & cost visibility** (`/cost`, TUI) | ❌ | ✅ Yes | ✅ Yes + Analytics |
| **Parallel Agents** | ❌ | ✅ 100+ | ✅ Unlimited | | **Parallel Agents** | ❌ | ✅ 100+ | ✅ Unlimited |
| **Support** | Community | ✅ Priority | ✅ Dedicated + Training | | **Support** | Community | ✅ Priority | ✅ Dedicated + Training |
| **Reporting** | Basic | ✅ Advanced | ✅ Custom Templates | | **Reporting** | Basic | ✅ Advanced | ✅ Custom Templates |
@ -368,7 +346,7 @@ Work with the CAI team to develop specialized capabilities.
- ✅ Unlimited alias1 tokens - ✅ Unlimited alias1 tokens
- ✅ Terminal UI with parallel agents - ✅ Terminal UI with parallel agents
- ✅ Context monitoring and optimization - ✅ Usage and cost visibility (`/cost`, compaction, TUI)
- ✅ Professional support (48h SLA) - ✅ Professional support (48h SLA)
- ✅ European data sovereignty (GDPR + NIS2) - ✅ European data sovereignty (GDPR + NIS2)
- ✅ Commercial use license - ✅ Commercial use license

6
docs/cai_pro_pricing.md
View File

@ -31,7 +31,7 @@ Free for research purposes.
- ❌ No alias1 model access - ❌ No alias1 model access
- ❌ No Terminal UI (TUI) - ❌ No Terminal UI (TUI)
- ❌ No parallel agent swarms - ❌ No parallel agent swarms
- ❌ No context monitoring - ❌ No PRO usage analytics and reporting
- ❌ No commercial license - ❌ No commercial license
- ⚠️ Framework updates ~6 months behind PRO - ⚠️ Framework updates ~6 months behind PRO
@ -49,7 +49,7 @@ Leading enterprise framework for AI Security with professional support.
- ✅ **Unlimited alias1 tokens** - ✅ **Unlimited alias1 tokens**
- ✅ **Terminal User Interface (TUI)** - ✅ **Terminal User Interface (TUI)**
- ✅ **Multi-agent parallel execution** (100+ agents) - ✅ **Multi-agent parallel execution** (100+ agents)
- ✅ **Context Monitoring** (`/context` command) - ✅ **Usage visibility** (`/cost`, TUI cost panels)
- ✅ **Commercial license** included - ✅ **Commercial license** included
- ✅ **Professional support** (48h SLA) - ✅ **Professional support** (48h SLA)
- ✅ **GDPR & NIS2 compliant** European hosting - ✅ **GDPR & NIS2 compliant** European hosting
@ -112,7 +112,7 @@ Custom deployment options.
| **🖥️ User Interfaces** | | | | | **🖥️ User Interfaces** | | | |
| Terminal UI (TUI) | ❌ | ✅ Multi-terminal | ✅ Multi-terminal + Custom | | Terminal UI (TUI) | ❌ | ✅ Multi-terminal | ✅ Multi-terminal + Custom |
| Parallel Agent Execution | ❌ | ✅ 100+ agents | ✅ Unlimited | | Parallel Agent Execution | ❌ | ✅ 100+ agents | ✅ Unlimited |
| Context Monitoring | ❌ | ✅ `/context` | ✅ + Analytics Dashboard | | Usage / cost visibility | ❌ | ✅ `/cost`, TUI | ✅ + Analytics Dashboard |
| Keyboard Shortcuts | ❌ | ✅ Full Set | ✅ Full Set + Custom | | Keyboard Shortcuts | ❌ | ✅ Full Set | ✅ Full Set + Custom |
| Team Presets | ❌ | ✅ 11 Teams | ✅ Unlimited Custom | | Team Presets | ❌ | ✅ 11 Teams | ✅ Unlimited Custom |
| | | | | | | | | |

14
docs/cai_pro_quickstart.md
View File

@ -228,18 +228,12 @@ CAI> Perform recon on https://bugbounty-target.com for a bug bounty program
## Advanced Configuration ## Advanced Configuration
### Enable Context Monitoring ### Check usage and costs
Track your token usage in real-time: Use **`/cost`** for session spend and token stats; use **`/compact`** when histories grow long.
```bash ```bash
CAI> /context CAI> /cost
# Shows:
# - Total tokens used/available
# - Breakdown by category (system, tools, memory, messages)
# - Visual grid representation
# - Optimization suggestions
``` ```
### Multi-Agent Parallel Execution ### Multi-Agent Parallel Execution
@ -364,7 +358,7 @@ Learn from real-world CAI applications:
### 🎯 Best Practices ### 🎯 Best Practices
1. **Start with clear prompts**: Be specific about your testing scope and objectives 1. **Start with clear prompts**: Be specific about your testing scope and objectives
2. **Use context monitoring**: Check `/context` regularly to optimize token usage 2. **Watch costs and thread length**: Use `/cost` and `/compact` as sessions grow
3. **Leverage parallel execution**: Run multiple agents for comprehensive coverage 3. **Leverage parallel execution**: Run multiple agents for comprehensive coverage
4. **Save your sessions**: Use `/save` to preserve important conversations 4. **Save your sessions**: Use `/save` to preserve important conversations
5. **Enable guardrails**: Keep `CAI_GUARDRAILS=true` for safer operations 5. **Enable guardrails**: Keep `CAI_GUARDRAILS=true` for safer operations

14
docs/cai_quickstart.md
View File

@ -106,17 +106,17 @@ CAI_PRICE_LIMIT="0.004" CAI_MODEL="qwen2.5:72b" cai
#### 3. Runtime configuration #### 3. Runtime configuration
After running CAI, use `/config` After running CAI, use **`/env`** (the catalog).
``` ```
/config set <number> <value> to configure a variable # see `config.py` or type `/help` /env set <#|NAME> <value...> # set a catalog variable
/env list # numbered catalog + live values
/env default # restore catalog defaults
``` ```
``` ```
cai cai
/config # It will display a panel with all the environment variables. /env list
# You must pick its reference NUMBER (1st column left) # Pick the catalog index or variable name from the first column
# `18` is the corresponding number for CAI_PRICE_LIMIT /env set 18 "0.004"
/config set 18 "0.004"
``` ```

28
docs/cli/advanced_usage.md
View File

@ -329,8 +329,8 @@ cai --prompt "/agent redteam_agent ; scan target.com ; /save results.json"
```text ```text
# ctf_workflow.txt # ctf_workflow.txt
/config CTF_NAME=hackableii /env set CTF_NAME hackableii
/config CTF_CHALLENGE=web_app /env set CTF_CHALLENGE web_app
/agent redteam_agent /agent redteam_agent
Analyze the CTF challenge environment Analyze the CTF challenge environment
Find and exploit vulnerabilities Find and exploit vulnerabilities
@ -343,7 +343,7 @@ Extract the flag
```text ```text
# bugbounty_recon.txt # bugbounty_recon.txt
/agent bug_bounter_agent /agent bug_bounter_agent
/config CAI_PRICE_LIMIT=20.0 /env set CAI_PRICE_LIMIT 20.0
# Reconnaissance # Reconnaissance
Perform subdomain enumeration on target.com Perform subdomain enumeration on target.com
@ -967,10 +967,10 @@ cai
CAI> /cost CAI> /cost
# Increase limit if needed # Increase limit if needed
CAI> /config CAI_PRICE_LIMIT=20.0 CAI> /env set CAI_PRICE_LIMIT 20.0
# Check updated limit # Check updated limit
CAI> /config | grep PRICE_LIMIT CAI> /env list | grep PRICE_LIMIT
``` ```
### Cost Optimization Strategies ### Cost Optimization Strategies
@ -1115,11 +1115,11 @@ cai
```bash ```bash
# Start with base config # Start with base config
CAI> /config CAI> /env list
# Adjust during session # Adjust during session
CAI> /config CAI_DEBUG=2 CAI> /env set CAI_DEBUG 2
CAI> /config CAI_PRICE_LIMIT=15.0 CAI> /env set CAI_PRICE_LIMIT 15.0
# Verify changes # Verify changes
CAI> /env | grep CAI CAI> /env | grep CAI
@ -1209,7 +1209,7 @@ Common issues and solutions.
CAI> /cost CAI> /cost
# Increase limit # Increase limit
CAI> /config CAI_PRICE_LIMIT=20.0 CAI> /env set CAI_PRICE_LIMIT 20.0
# Or restart with higher limit # Or restart with higher limit
exit exit
@ -1223,7 +1223,7 @@ CAI_PRICE_LIMIT=20.0 cai
CAI> /env | grep MAX_INTERACTIONS CAI> /env | grep MAX_INTERACTIONS
# Increase limit # Increase limit
CAI> /config CAI_MAX_INTERACTIONS=500 CAI> /env set CAI_MAX_INTERACTIONS 500
# Or use /flush to start fresh # Or use /flush to start fresh
CAI> /flush CAI> /flush
@ -1242,14 +1242,14 @@ CAI> /agent
CAI> /agent redteam_agent CAI> /agent redteam_agent
# Check configuration # Check configuration
CAI> /config CAI> /env list
``` ```
### Issue: Context Window Full ### Issue: Context Window Full
```bash ```bash
# Check context usage (CAI PRO) # Review spend / token usage
CAI> /context CAI> /cost
# Compact conversation # Compact conversation
CAI> /compact CAI> /compact
@ -1298,7 +1298,7 @@ export CAI_DEBUG=2
cai cai
# Or enable during session # Or enable during session
CAI> /config CAI_DEBUG=2 CAI> /env set CAI_DEBUG 2
``` ```
--- ---

14
docs/cli/cli_index.md
View File

@ -92,11 +92,11 @@ See the [Getting Started Guide](getting_started.md) for detailed instructions.
Over 30 built-in commands organized by category: Over 30 built-in commands organized by category:
- **Agent Management**: `/agent`, `/parallel`, `/run` - **Agent Management**: `/agent`, `/parallel`
- **Memory & History**: `/memory`, `/history`, `/compact`, `/flush`, `/load`, `/merge` - **Memory & History**: `/memory`, `/history`, `/compact`, `/flush`, `/load`, `/merge`, `/save`
- **Environment & Config**: `/config`, `/env`, `/help var` (per-variable help), `/workspace`, `/virtualization` - **Environment**: `/env` (catalog `list` / `get` / `set` / `default`), `/help var` (per-variable help), `/workspace`, `/virtualization`
- **Tools & Integration**: `/mcp`, `/platform`, `/shell` - **Tools & Integration**: `/mcp`, `/shell`
- **Utilities**: `/model`, `/graph`, `/context`, `/cost`, `/help` - **Utilities**: `/model`, `/graph`, `/cost`, `/help`
All commands support aliases for faster typing (e.g., `/a` for `/agent`, `/h` for `/help`). All commands support aliases for faster typing (e.g., `/a` for `/agent`, `/h` for `/help`).
@ -280,7 +280,7 @@ cai --yaml agents.yaml --prompt "full reconnaissance on target.com"
| `/agent list` | List all agents | `/agent list` | | `/agent list` | List all agents | `/agent list` |
| `/agent <name>` | Switch agent | `/agent redteam_agent` | | `/agent <name>` | Switch agent | `/agent redteam_agent` |
| `/model <name>` | Change model | `/model alias1` | | `/model <name>` | Change model | `/model alias1` |
| `/config` | View configuration | `/config` | | `/env list` | Catalog + live values | `/env list` |
| `/help` | Show help | `/help agent` | | `/help` | Show help | `/help agent` |
| `/save <file>` | Save session | `/save session.json` | | `/save <file>` | Save session | `/save session.json` |
| `/load <file>` | Load session | `/load session.json` | | `/load <file>` | Load session | `/load session.json` |
@ -306,7 +306,7 @@ CAI CLI can be configured via:
1. **Environment Variables**: `CAI_MODEL`, `CAI_AGENT_TYPE`, etc. 1. **Environment Variables**: `CAI_MODEL`, `CAI_AGENT_TYPE`, etc.
2. **`.env` File**: Place in your working directory 2. **`.env` File**: Place in your working directory
3. **`/config` Command**: Runtime configuration changes 3. **`/env` command**: Runtime changes via catalog (`/env set …`, `/env default`, …)
4. **YAML Files**: Agent and workflow definitions 4. **YAML Files**: Agent and workflow definitions
Example `.env`: Example `.env`:

1907
docs/cli/commands_reference.md
View File

File diff suppressed because it is too large Load Diff

45
docs/cli/getting_started.md
View File

@ -71,18 +71,18 @@ export ALIAS_API_KEY="ak_live_1234567890abcdef"
cai cai
``` ```
### Method 3: Runtime Configuration ### Method 3: Runtime configuration
After launching CAI, use the `/config` command: After launching CAI, use the **`/env`** catalog:
```bash ```bash
CAI> /config CAI_MODEL=alias1 CAI> /env set CAI_MODEL alias1
``` ```
To view all current configuration: To view the catalog with live values:
```bash ```bash
CAI> /config CAI> /env list
``` ```
## Step 3: Select Your Model ## Step 3: Select Your Model
@ -96,7 +96,7 @@ CAI> /model alias1
To see all available models: To see all available models:
```bash ```bash
CAI> /model-show CAI> /model show
``` ```
### Recommended Models ### Recommended Models
@ -190,8 +190,8 @@ The agent will:
```bash ```bash
# Set up CTF environment # Set up CTF environment
CAI> /config CTF_NAME=hackableii CAI> /env set CTF_NAME hackableii
CAI> /config CTF_CHALLENGE=web_challenge CAI> /env set CTF_CHALLENGE web_challenge
# Start the challenge # Start the challenge
CAI> Analyze this CTF challenge and find the flag CAI> Analyze this CTF challenge and find the flag
@ -249,8 +249,8 @@ CAI> /model
# Change model # Change model
CAI> /model gpt-4o CAI> /model gpt-4o
# List all available models # List all available models (LiteLLM catalog)
CAI> /model-show CAI> /model show
``` ```
### Session Management ### Session Management
@ -322,23 +322,23 @@ CAI> Run a comprehensive port scan on 192.168.1.0/24
## Step 8: Working with Configuration ## Step 8: Working with Configuration
### View Current Configuration ### View current configuration
```bash ```bash
CAI> /config CAI> /env list
``` ```
This displays a panel with all environment variables and their current values. This shows the environment catalog with numbers and current values. Use bare **`/env`** for a compact view of `CAI_*` / `CTF_*` only.
### Change Configuration at Runtime ### Change configuration at runtime
```bash ```bash
# Set a specific variable (use the number from /config output) # Set by catalog index (from /env list)
CAI> /config set 18 "5.0" CAI> /env set 18 "5.0"
# Or set by name # Or set by variable name
CAI> /config CAI_PRICE_LIMIT=5.0 CAI> /env set CAI_PRICE_LIMIT 5.0
CAI> /config CAI_MAX_TURNS=50 CAI> /env set CAI_MAX_TURNS 50
``` ```
### Important Configuration Variables ### Important Configuration Variables
@ -483,7 +483,7 @@ CAI> /help <command_name>
CAI> /cost CAI> /cost
# Increase limit # Increase limit
CAI> /config CAI_PRICE_LIMIT=20.0 CAI> /env set CAI_PRICE_LIMIT 20.0
# Or set it before launching # Or set it before launching
CAI_PRICE_LIMIT=20.0 cai CAI_PRICE_LIMIT=20.0 cai
@ -494,7 +494,7 @@ CAI_PRICE_LIMIT=20.0 cai
**Solution**: **Solution**:
```bash ```bash
# Increase turn limit # Increase turn limit
CAI> /config CAI_MAX_TURNS=100 CAI> /env set CAI_MAX_TURNS 100
# Or flush history and start fresh # Or flush history and start fresh
CAI> /flush CAI> /flush
@ -533,7 +533,8 @@ Congratulations! You've completed the basics of CAI CLI. Here's what to explore
/agent list # List all agents /agent list # List all agents
/agent <name> # Switch agent /agent <name> # Switch agent
/model <name> # Change model /model <name> # Change model
/config # View configuration /model show # LiteLLM model catalog
/env list # Catalog + live values
/help # Get help /help # Get help
/save <file> # Save session /save <file> # Save session
/load <file> # Load session /load <file> # Load session

295
docs/council.md
View File

@ -1,295 +0,0 @@
# LLM Council
The LLM Council is a multi-model consensus system that queries multiple AI models, has them evaluate each other's responses, and synthesizes a final answer through a chairman. This approach improves accuracy for complex decisions by leveraging diverse model perspectives.
## How It Works
The council operates in three stages:
```
User Query: "What's the best approach for this security task?"
│
├─ Stage 1: Council Members (parallel)
│ ├─ Agent + gpt-4o → Text response only
│ ├─ Agent + gpt-5 → Text response only
│ └─ Agent + claude-sonnet-4-5 → Text response only
│
├─ Stage 2: Rankings (parallel)
│ ├─ Agent + gpt-4o → Ranks anonymized responses
│ ├─ Agent + gpt-5 → Ranks anonymized responses
│ └─ Agent + claude-sonnet-4-5 → Ranks anonymized responses
│
└─ Stage 3: Chairman
└─ Active Agent + TOOLS
├─ Synthesizes best answer
└─ Can execute operations if requested
```
**Key Points:**
- **Stage 1 & 2**: Council members provide text-only responses (no tool execution)
- **Stage 3**: The chairman (your active agent) can use all available tools
- All members use the current agent's instructions and context
## Quick Start
```bash
# Configure council models
export CAI_COUNCIL="gpt-4o,gpt-5,claude-sonnet-4-20250514"
# In CAI REPL, load an agent first
CAI> /agent redteam
# Use the council command
CAI> /council What are the best practices for API security?
# Or use the short alias
CAI> /c How should I approach this vulnerability assessment?
```
## Configuration
### Environment Variables
| Variable | Description | Default |
|----------|-------------|---------|
| `CAI_COUNCIL` | Comma-separated list of council member models | `gpt-4o,gpt-4o-mini` |
| `CAI_COUNCIL_AUTO` | Auto-convene setting: `false`, `true`/`1`, or interval number | `false` |
| `CAI_COUNCIL_PROMPT` | Custom prompt for auto-council reviews | See below |
| `CAI_COUNCIL_DEBUG` | Enable debug output (`1`, `true`, `yes`) | `false` |
```bash
# Example configuration
export CAI_COUNCIL="gpt-4o,gpt-5,claude-sonnet-4-20250514"
export CAI_COUNCIL_AUTO="5"
export CAI_COUNCIL_PROMPT="Review the current progress and recommend the best approach."
export CAI_COUNCIL_DEBUG="1"
```
### API Keys
Ensure you have the appropriate API keys set for your council models:
```bash
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY="sk-ant-..."
export ALIAS_API_KEY="..."
```
### Verified Model Names
Use exact model names as shown in the `/model` command:
| Provider | Models |
|----------|--------|
| OpenAI | `gpt-5`, `gpt-4o`, `gpt-4o-mini`, `o3-mini` |
| Anthropic | `claude-sonnet-4-20250514`, `claude-3-5-sonnet-20240620` |
| Alias | `alias1` |
| DeepSeek | `deepseek-v3`, `deepseek-r1` |
## Manual Usage
The `/council` command (alias `/c`) invokes the council manually:
```bash
# Load an agent
CAI> /agent redteam
# Ask the council
CAI> /council What vulnerabilities should I look for in this web application?
```
The council uses the active agent's:
- Instructions/system prompt
- Available tools (chairman only)
- Guardrails
## Auto-Council Mode
When `CAI_COUNCIL_AUTO` is enabled, the council convenes automatically at specified intervals during agent execution.
### Configuration Options
- `false` - Never auto-convene (use `/council` manually)
- `true` or `1` - Convene at every agent interaction
- `5`, `10`, etc. - Convene every N interactions
### Example: Every Interaction
```bash
export CAI_COUNCIL_AUTO="1"
CAI> run ps aux, then analyze the results, then check for vulnerabilities
🏛️ COUNCIL (auto-invoked at interaction [1])
[Stage 1, 2, 3 run...]
[1] Agent: "I'll run ps aux" → executes command
🏛️ COUNCIL (auto-invoked at interaction [2])
[Stage 1, 2, 3 run...]
[2] Agent: "Analyzing results..." → analyzes output
🏛️ COUNCIL (auto-invoked at interaction [3])
[Stage 1, 2, 3 run...]
[3] Agent: "Checking for vulnerabilities..." → performs check
```
### Example: Every 5 Interactions
```bash
export CAI_COUNCIL_AUTO="5"
CAI> perform a comprehensive security audit
[1] Agent executes first task
[2] Agent executes second task
[3] Agent executes third task
[4] Agent executes fourth task
🏛️ COUNCIL (auto-invoked at interaction [5])
[Stage 1, 2, 3 run...]
[5] Agent executes fifth task
[6] Agent continues...
[7] Agent continues...
[8] Agent continues...
[9] Agent continues...
🏛️ COUNCIL (auto-invoked at interaction [10])
[Stage 1, 2, 3 run...]
[10] Agent executes tenth task
```
## Programmatic Usage
You can use the council directly in Python code:
```python
from cai.council import run_full_council_agents, CouncilAgentConfig
from cai.sdk.agents import Agent
# With an existing agent
stage1, stage2, stage3, metadata = await run_full_council_agents(
base_agent=my_agent,
user_query="Your question here",
)
# Access results
print(stage3["response"]) # Final answer
print(metadata["aggregate_rankings"]) # Model rankings
print(metadata["council_cost"]) # Total cost
print(metadata["council_input_tokens"]) # Input tokens
print(metadata["council_output_tokens"]) # Output tokens
```
### Return Values
```python
stage1_results: List[Dict] # Individual responses from each model
stage2_results: List[Dict] # Rankings from each model
stage3_result: Dict # Final synthesized answer
metadata: Dict # Rankings, cost, tokens
```
### Metadata Structure
```python
metadata = {
"aggregate_rankings": [
{"model": "gpt-4o", "average_rank": 1.33, "rankings_count": 3},
{"model": "gpt-5", "average_rank": 2.0, "rankings_count": 3},
],
"council_cost": 0.032,
"council_input_tokens": 5000,
"council_output_tokens": 2500,
}
```
## Visual Display
During execution, the council shows an animated panel with progress:
```
╭─────────────────────── Alias Council ────────────────────────╮
│ │
│ 👑 Chairman: Red Team Agent (gpt-4o) │
│ │
│ ⠋ Stage 1: Collecting responses from council members │
│ ██████████░░░░░░░░░░ 2/3 │
│ ✓ gpt-4o │
│ ✓ gpt-5 │
│ ⠋ alias1 │
│ │
│ ○ Stage 2: Waiting... │
│ │
│ 💰 $0.012 (1.2k in / 800 out) ⏱ 15.2s │
│ │
╰─────────────────────────────────────────────────────────────────╯
```
## Performance Considerations
| Metric | Single Query | Council (3 models) |
|--------|--------------|-------------------|
| API Calls | 1 | ~7 (2N + 1) |
| Cost | 1x | 3-4x |
| Latency | 1x | 2-3x |
| Accuracy | Base | Improved |
!!! tip "When to Use Council"
Use the council when accuracy matters more than speed or cost. It's particularly valuable for:
- Complex security decisions
- Architecture recommendations
- Vulnerability assessments
- Strategic planning tasks
## Troubleshooting
### Debug Mode
Enable detailed logging to diagnose issues:
```bash
export CAI_COUNCIL_DEBUG=1
```
### Common Issues
**"All models failed to respond"**
- Verify API keys are set correctly
- Check model names with `/model` command
- Check for rate limiting
**Council hangs on Stage 1**
- Model name might be incorrect (verify with `/model`)
- API key invalid or missing
- Network connectivity issues
**"Temperature not supported"**
- Handled automatically for GPT-5/O1/O3 models (temperature set to 1)
### Test Individual Models
Before using council, verify each model works independently:
```bash
CAI> /model gpt-4o
CAI> What is 2+2?
```
### Minimal Configuration
If experiencing issues, try a minimal setup:
```bash
export CAI_COUNCIL="gpt-4o,gpt-4o-mini"
```
## Credits
Inspired by [llm-council](https://github.com/karpathy/llm-council) by Andrej Karpathy.

11
docs/environment_variables.md
View File

@ -10,9 +10,10 @@ In current CAI releases, you can explore environment variables **from inside the
| What you need | Command | | What you need | Command |
|---------------|---------| |---------------|---------|
| **Numbered list with live values** (what is set *now*) | `/config` or `/config list` | | **Numbered catalog with live values** (what is set *now*) | `/env list` |
| **Session keys only** (`CAI_*` / `CTF_*` in this process) | `/env` (no arguments) |
| **Full reference tables** (defaults, allowed values, when they apply, extras) | `/help` — scroll past the quick guide; or `/help topics` for the overview first, then the same tables at the end | | **Full reference tables** (defaults, allowed values, when they apply, extras) | `/help` — scroll past the quick guide; or `/help topics` for the overview first, then the same tables at the end |
| **Long-form help for one variable** (examples, `/config` index when listed, notes) | `/help var VARIABLE_NAME` (e.g. `/help var CAI_MODEL`, `/help var CAI_AVOID_SUDO`) | | **Long-form help for one variable** (examples, catalog index when listed, notes) | `/help var VARIABLE_NAME` (e.g. `/help var CAI_MODEL`, `/help var CAI_AVOID_SUDO`) |
Aliases such as `/h` for `/help` work the same way. This page remains the **canonical web reference**; the REPL output tracks the version you have installed. Aliases such as `/h` for `/help` work the same way. This page remains the **canonical web reference**; the REPL output tracks the version you have installed.
@ -32,11 +33,11 @@ In the REPL, **`/help var VARIABLE_NAME`** expands each variable with the same i
**How to set (matches `/help` copy):** **How to set (matches `/help` copy):**
- Before launch: `export VAR=value` or a line in `.env`, then start CAI. - Before launch: `export VAR=value` or a line in `.env`, then start CAI.
- During a session: `/config set <number> <value>`, `/config VAR=value`, or code updating `os.environ`. - During a session: `/env set <#|NAME> <value…>`, `/env default`, or code updating `os.environ`.
**Types (short):** *bool* — `true`/`false`, `1`/`0`, etc.; *string* — free text; *int* / *float* — numeric; ranges in **Values** are the usual bounds CAI documents; *secret* — treat like a string, never commit real keys. **Types (short):** *bool* — `true`/`false`, `1`/`0`, etc.; *string* — free text; *int* / *float* — numeric; ranges in **Values** are the usual bounds CAI documents; *secret* — treat like a string, never commit real keys.
For **numbered `/config` index**, **extra notes**, and **copy-paste examples** per variable, use **`/help var NAME`** in the REPL — the web page keeps one compact table for browsing. For **numbered catalog index**, **extra notes**, and **copy-paste examples** per variable, use **`/help var NAME`** in the REPL — the web page keeps one compact table for browsing.
--- ---
@ -398,5 +399,5 @@ CAI_MODEL="gpt-4o" CAI_PRICE_LIMIT="2" cai
**3. Runtime configuration** **3. Runtime configuration**
Use slash commands during a session: `/config`, `/config set …`, and the in-session help above (`/help`, `/help var …`). See [Commands Reference](cai/getting-started/commands.md) and [CLI Commands Reference](cli/commands_reference.md). Use slash commands during a session: `/env`, `/env list`, `/env set …`, and the in-session help above (`/help`, `/help var …`). See [Commands Reference](cai/getting-started/commands.md) and [CLI Commands Reference](cli/commands_reference.md).

2
docs/index.md
View File

@ -35,7 +35,7 @@ A lightweight, ergonomic framework for building bug bounty-ready Cybersecurity A
- ✅ **Unlimited `alias1` tokens** - Our state-of-the-art cybersecurity model that beats GPT-5 in benchmarks - ✅ **Unlimited `alias1` tokens** - Our state-of-the-art cybersecurity model that beats GPT-5 in benchmarks
- ✅ **Mobile User Interface (iOS)** - Native iOS app for security testing on the go - **[Join TestFlight Beta](https://testflight.apple.com/join/nXZZD4Z5)** - ✅ **Mobile User Interface (iOS)** - Native iOS app for security testing on the go - **[Join TestFlight Beta](https://testflight.apple.com/join/nXZZD4Z5)**
- ✅ **Terminal User Interface (TUI)** - Multi-agent parallel execution with visual monitoring *(Deprecated - Use Mobile UI)* - ✅ **Terminal User Interface (TUI)** - Multi-agent parallel execution with visual monitoring *(Deprecated - Use Mobile UI)*
- ✅ **Context Monitoring** - Real-time token tracking and optimization - ✅ **Usage visibility** - `/cost`, compaction, and TUI cost indicators
- ✅ **Zero Refusals** - Unrestricted AI specifically trained for offensive security - ✅ **Zero Refusals** - Unrestricted AI specifically trained for offensive security
- ✅ **European Hosting** - GDPR & NIS2 compliant with guaranteed data privacy - ✅ **European Hosting** - GDPR & NIS2 compliant with guaranteed data privacy
- ✅ **Professional Support** - Dedicated technical assistance from security experts - ✅ **Professional Support** - Dedicated technical assistance from security experts

4
docs/providers/ollama_cloud.md
View File

@ -25,7 +25,7 @@ cai
## Available Models ## Available Models
View in CAI with `/model-show` under "Ollama Cloud" category: View in CAI with `/model show` under "Ollama Cloud" category:
- `ollama_cloud/gpt-oss:120b` - General purpose 120B model - `ollama_cloud/gpt-oss:120b` - General purpose 120B model
- `ollama_cloud/llama3.3:70b` - Llama 3.3 70B - `ollama_cloud/llama3.3:70b` - Llama 3.3 70B
@ -40,7 +40,7 @@ More models at [ollama.com/library](https://ollama.com/library).
# By name # By name
CAI> /model ollama_cloud/gpt-oss:120b CAI> /model ollama_cloud/gpt-oss:120b
# By number (after /model-show) # By number (after /model show)
CAI> /model 3 CAI> /model 3
``` ```

178
docs/tui/commands_reference.md
View File

@ -5,6 +5,9 @@
--- ---
!!! tip "Recommended: use the CLI REPL"
The command matrix for every `cai` binary flag and REPL slash command is maintained in the **[CLI commands reference](../cli/commands_reference.md)**. **When in doubt—especially for scripting, copy-paste examples, or comparing subcommands—use the plain CLI REPL** and the CLI reference. The TUI shares the same underlying engine; this document focuses on **multi-terminal UX**, keyboard shortcuts, palette actions, and behaviors that differ from the CLI.
This comprehensive guide documents all commands available in the CAI Terminal User Interface (TUI), including command palette actions, keyboard shortcuts, and CLI-style commands. This comprehensive guide documents all commands available in the CAI Terminal User Interface (TUI), including command palette actions, keyboard shortcuts, and CLI-style commands.
--- ---
@ -23,19 +26,40 @@ CAI TUI commands are organized into the following categories:
--- ---
## CLI parity and TUI-only differences
The [CLI commands reference](../cli/commands_reference.md) is the **authoritative** list for REPL slash commands. The TUI reuses the same slash commands, but a few interactions differ:
| Situation | CLI REPL | TUI |
| ------------------------------------------ | -------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------ |
| `/flush` or `/clear` with **no** arguments | Interactive menu of agents that still have history | Clears history for the **focused terminal only** |
| `/agent` with no arguments | Same as `/agent current` | Shows the active agent configuration across terminals where applicable |
| Screen vs history | N/A | **Ctrl+L** / visual “clear” only affects scrollback; it does **not** replace `/flush` for wiping message history |
For binary flags such as `--tui`, `--yaml`, or `--api`, see [Binary `cai` CLI flags](../cli/commands_reference.md#binary-cai-cli-flags).
---
## Agent Management ## Agent Management
### `/agent` or `/a` ### `/agent` or `/a`
Switch between agents or list all available agents. Switch between agents or list all available agents.
!!! note "CLI reference vs TUI examples"
The CLI reference documents `/agent` with explicit subcommands (`list`, `select`, `info`, `current`, …). In the TUI you will often see the shorthand `/agent <name>`; on many builds that is equivalent to `/agent select <name>`. For exact syntax and flags, follow the **[CLI commands reference](../cli/commands_reference.md#agent-a)**.
**Syntax**: **Syntax**:
``` ```
/agent [agent_name] /agent [agent_name]
/a [agent_name] /a [agent_name]
``` ```
**Examples**: **Examples**:
```bash ```bash
# List all available agents # List all available agents
/agent /agent
@ -48,6 +72,7 @@ Switch between agents or list all available agents.
``` ```
**Available Agents**: **Available Agents**:
- `redteam_agent` - Offensive security testing and penetration testing - `redteam_agent` - Offensive security testing and penetration testing
- `blueteam_agent` - Defensive security analysis and hardening - `blueteam_agent` - Defensive security analysis and hardening
- `bug_bounter_agent` - Bug bounty hunting and vulnerability research - `bug_bounter_agent` - Bug bounty hunting and vulnerability research
@ -72,6 +97,7 @@ Switch between agents or list all available agents.
- `offsec_pattern` - Offensive security pattern orchestration - `offsec_pattern` - Offensive security pattern orchestration
**Notes**: **Notes**:
- Agent changes are immediate and affect only the active terminal - Agent changes are immediate and affect only the active terminal
- Each terminal can run a different agent simultaneously - Each terminal can run a different agent simultaneously
- Agent context is preserved when switching between terminals - Agent context is preserved when switching between terminals
@ -85,6 +111,7 @@ Switch between agents or list all available agents.
CAI TUI uses model dropdowns in each terminal header for model management. Models are configured via environment variables and aliases. CAI TUI uses model dropdowns in each terminal header for model management. Models are configured via environment variables and aliases.
**Available Models**: **Available Models**:
- `alias1` - Cybersecurity focus model [Recommended] - `alias1` - Cybersecurity focus model [Recommended]
- `gpt-4o` - OpenAI GPT-4 Optimized - `gpt-4o` - OpenAI GPT-4 Optimized
- `gpt-4-turbo` - OpenAI GPT-4 Turbo - `gpt-4-turbo` - OpenAI GPT-4 Turbo
@ -93,11 +120,13 @@ CAI TUI uses model dropdowns in each terminal header for model management. Model
- `o1-preview` - OpenAI O1 Preview - `o1-preview` - OpenAI O1 Preview
**How to Change Models**: **How to Change Models**:
1. Click the model dropdown in any terminal header 1. Click the model dropdown in any terminal header
2. Select desired model from the list 2. Select desired model from the list
3. Model change takes effect immediately for that terminal 3. Model change takes effect immediately for that terminal
**Environment Variables**: **Environment Variables**:
```bash ```bash
export CAI_MODEL=gpt-4o # Set default model export CAI_MODEL=gpt-4o # Set default model
export CAI_OPENAI_API_KEY=sk-... # OpenAI API key export CAI_OPENAI_API_KEY=sk-... # OpenAI API key
@ -105,6 +134,7 @@ export CAI_ANTHROPIC_API_KEY=sk-... # Anthropic API key
``` ```
**Notes**: **Notes**:
- Each terminal can use a different model - Each terminal can use a different model
- Model costs are tracked separately per terminal - Model costs are tracked separately per terminal
- Switching models mid-conversation preserves history - Switching models mid-conversation preserves history
@ -120,11 +150,13 @@ Send commands to specific terminals using either the prefix notation or the flag
#### Method 1: Prefix Notation #### Method 1: Prefix Notation
**Syntax**: **Syntax**:
``` ```
T<terminal_number>:<command> T<terminal_number>:<command>
``` ```
**Examples**: **Examples**:
```bash ```bash
# Switch agent in Terminal 2 # Switch agent in Terminal 2
T2:/agent blueteam_agent T2:/agent blueteam_agent
@ -138,6 +170,7 @@ T1:/clear
# Execute command in Terminal 4 # Execute command in Terminal 4
T4:scan target.com for vulnerabilities T4:scan target.com for vulnerabilities
``` ```
#### Method 2: Flag Notation #### Method 2: Flag Notation
**Syntax**: **Syntax**:
@ -148,6 +181,7 @@ T4:scan target.com for vulnerabilities
``` ```
**Examples**: **Examples**:
```bash ```bash
# Switch agent in Terminal 2 # Switch agent in Terminal 2
/agent blueteam_agent t2 /agent blueteam_agent t2
@ -166,6 +200,7 @@ Scan target.com for XSS vulnerabilities t2
``` ```
**Supported Flags**: **Supported Flags**:
- `t1` - Target Terminal 1 - `t1` - Target Terminal 1
- `t2` - Target Terminal 2 - `t2` - Target Terminal 2
- `t3` - Target Terminal 3 - `t3` - Target Terminal 3
@ -173,6 +208,7 @@ Scan target.com for XSS vulnerabilities t2
- (Additional terminals if configured: `t5`, `t6`, etc.) - (Additional terminals if configured: `t5`, `t6`, etc.)
**Notes**: **Notes**:
- Both methods achieve the same result - Both methods achieve the same result
- Flag notation is more concise for quick commands - Flag notation is more concise for quick commands
- Prefix notation is clearer for complex prompts - Prefix notation is clearer for complex prompts
@ -183,6 +219,7 @@ Scan target.com for XSS vulnerabilities t2
**Keyboard Shortcut**: Click the `[+]` button in the top bar **Keyboard Shortcut**: Click the `[+]` button in the top bar
**Notes**: **Notes**:
- New terminals start with `redteam_agent` by default - New terminals start with `redteam_agent` by default
- Maximum recommended terminals: 4 (for optimal UX) - Maximum recommended terminals: 4 (for optimal UX)
- Terminals beyond 4 use scrollable layout - Terminals beyond 4 use scrollable layout
@ -191,16 +228,18 @@ Scan target.com for XSS vulnerabilities t2
## History and Memory ## History and Memory
### `/history [number] [agent_name]` or `/h` ### `/history [number] [agent_name]` or `/his`
Display conversation history for the current or specified agent. Display conversation history for the current or specified agent. (Use `/his` — `/h` is reserved for `/help` on builds that follow the documented alias table.)
**Syntax**: **Syntax**:
``` ```
/history [number] [agent_name] /history [number] [agent_name]
``` ```
**Examples**: **Examples**:
```bash ```bash
# Show last 10 messages # Show last 10 messages
/history /history
@ -212,24 +251,27 @@ Display conversation history for the current or specified agent.
/history 10 redteam_agent /history 10 redteam_agent
# Compact syntax # Compact syntax
/h 5 /his 5
``` ```
**Notes**: **Notes**:
- Default shows last 10 interactions - Default shows last 10 interactions
- History includes both user prompts and agent responses - History includes both user prompts and agent responses
- History is terminal-specific - History is terminal-specific
### `/flush [agent_name|all]` ### `/flush [agent_name|all]` or `/clear`
Clear agent message history. Clear agent message history (`/clear` is an alias of `/flush` for history, not the same as clearing the screen—see below).
**Syntax**: **Syntax**:
``` ```
/flush [agent_name|all] /flush [agent_name|all]
``` ```
**Examples**: **Examples**:
```bash ```bash
# Flush current agent history # Flush current agent history
/flush /flush
@ -242,15 +284,18 @@ Clear agent message history.
``` ```
**Notes**: **Notes**:
- Flushing is irreversible - Flushing is irreversible
- Agent context window is reset - Agent context window is reset
- Useful for starting fresh conversations - Useful for starting fresh conversations
- **TUI:** with **no** arguments, `/flush` typically affects only the **focused** terminal (see [CLI parity and TUI-only differences](#cli-parity-and-tui-only-differences)). The CLI may instead show a picker—confirm with `/help flush` on your build.
### `/memory [subcommand]` or `/mem` ### `/memory [subcommand]` or `/mem`
Advanced memory management for agents. Advanced memory management for agents.
**Syntax**: **Syntax**:
``` ```
/memory <subcommand> /memory <subcommand>
/mem <subcommand> /mem <subcommand>
@ -259,55 +304,72 @@ Advanced memory management for agents.
**Subcommands**: **Subcommands**:
#### `list` #### `list`
Show all saved memories. Show all saved memories.
```bash ```bash
/memory list /memory list
``` ```
#### `save [name]` #### `save [name]`
Save current conversation as a memory. Save current conversation as a memory.
```bash ```bash
/memory save "Authentication bypass research" /memory save "Authentication bypass research"
/mem save pentest_findings /mem save pentest_findings
``` ```
#### `apply <memory_id>` #### `apply <memory_id>`
Apply a saved memory to the current agent. Apply a saved memory to the current agent.
```bash ```bash
/memory apply mem_12345 /memory apply mem_12345
``` ```
#### `show <memory_id>` #### `show <memory_id>`
Display the content of a specific memory. Display the content of a specific memory.
```bash ```bash
/memory show mem_12345 /memory show mem_12345
``` ```
#### `delete <memory_id>` #### `delete <memory_id>`
Remove a memory permanently. Remove a memory permanently.
```bash ```bash
/memory delete mem_12345 /memory delete mem_12345
``` ```
#### `merge <id1> <id2> [name]` #### `merge <id1> <id2> [name]`
Combine two memories into one. Combine two memories into one.
```bash ```bash
/memory merge mem_12345 mem_67890 "Combined pentesting notes" /memory merge mem_12345 mem_67890 "Combined pentesting notes"
``` ```
#### `compact` #### `compact`
AI-powered memory summarization. AI-powered memory summarization.
```bash ```bash
/memory compact /memory compact
``` ```
#### `status` #### `status`
Show memory system status and statistics. Show memory system status and statistics.
```bash ```bash
/memory status /memory status
``` ```
**Notes**: **Notes**:
- Memories persist across sessions - Memories persist across sessions
- Useful for resuming long-term research projects - Useful for resuming long-term research projects
- AI-powered summarization reduces token usage - AI-powered summarization reduces token usage
@ -321,15 +383,18 @@ Show memory system status and statistics.
Save the current conversation to a file. Save the current conversation to a file.
**Syntax**: **Syntax**:
``` ```
/save <filename> /save <filename>
``` ```
**Supported Formats**: **Supported Formats**:
- JSON (`.json`) - JSON (`.json`)
- Markdown (`.md`) - Markdown (`.md`)
**Examples**: **Examples**:
```bash ```bash
# Save as JSON # Save as JSON
/save pentest_session.json /save pentest_session.json
@ -342,6 +407,7 @@ Save the current conversation to a file.
``` ```
**Notes**: **Notes**:
- Saves all terminal conversations - Saves all terminal conversations
- Includes agent names, models, and timestamps - Includes agent names, models, and timestamps
- Cost information is preserved - Cost information is preserved
@ -351,12 +417,14 @@ Save the current conversation to a file.
Load a previously saved conversation. Load a previously saved conversation.
**Syntax**: **Syntax**:
``` ```
/load <filename> /load <filename>
/l <filename> /l <filename>
``` ```
**Examples**: **Examples**:
```bash ```bash
# Load JSON session # Load JSON session
/load pentest_session.json /load pentest_session.json
@ -369,6 +437,7 @@ Load a previously saved conversation.
``` ```
**Notes**: **Notes**:
- Restores agent context and history - Restores agent context and history
- Compatible with both JSON and Markdown formats - Compatible with both JSON and Markdown formats
- Loading does not affect current cost tracking - Loading does not affect current cost tracking
@ -377,61 +446,20 @@ Load a previously saved conversation.
## Utility Commands ## Utility Commands
### `/context [agent_name]` or `/ctx` 🚀 **CAI PRO Exclusive** For spend and token usage use **`/cost`** (and the per-terminal cost UI in the TUI). When threads grow too long, use **`/compact`**. See the [CLI commands reference](../cli/commands_reference.md) for the full command list.
> **⚡ CAI PRO Exclusive Feature**
> The `/context` command is available exclusively in **[CAI PRO](https://aliasrobotics.com/cybersecurityai.php)**. To access this feature and unlock advanced monitoring capabilities, visit [Alias Robotics](https://aliasrobotics.com/cybersecurityai.php) for more information.
Display context window usage and token statistics for the current conversation.
**Syntax**:
```
/context [agent_name]
/ctx [agent_name]
```
**Examples**:
```bash
# Show context usage for active terminal
/context
# Show context usage for specific agent
/context redteam_agent
# Compact syntax
/ctx
```
**Output Includes**:
- Total context usage (used/max tokens) with percentage
- Visual grid representation with CAI logo
- Breakdown by category:
- System prompt tokens
- Tool definitions tokens
- Memory files (RAG) tokens
- User prompts tokens
- Assistant responses tokens
- Tool calls tokens
- Tool results tokens
- Free space available
- Color-coded visualization for easy identification
**Notes**:
- Context usage helps monitor when you're approaching model limits
- Different models have different context windows (e.g., GPT-4: 128k, Claude: 200k)
- Use `/context` regularly during long conversations to avoid hitting limits
- Context usage is terminal-specific in TUI mode
### `/cost [agent_name]` ### `/cost [agent_name]`
Display API usage costs and token statistics. Display API usage costs and token statistics.
**Syntax**: **Syntax**:
``` ```
/cost [agent_name] /cost [agent_name]
``` ```
**Examples**: **Examples**:
```bash ```bash
# Show costs for active terminal # Show costs for active terminal
/cost /cost
@ -444,6 +472,7 @@ Display API usage costs and token statistics.
``` ```
**Output Includes**: **Output Includes**:
- Total cost (USD) - Total cost (USD)
- Input tokens used - Input tokens used
- Output tokens used - Output tokens used
@ -456,12 +485,14 @@ Display API usage costs and token statistics.
Get help for commands. Get help for commands.
**Syntax**: **Syntax**:
``` ```
/help [command] /help [command]
/? [command] /? [command]
``` ```
**Examples**: **Examples**:
```bash ```bash
# General help # General help
/help /help
@ -477,11 +508,13 @@ Get help for commands.
Display environment variables relevant to CAI. Display environment variables relevant to CAI.
**Syntax**: **Syntax**:
``` ```
/env /env
``` ```
**Output Includes**: **Output Includes**:
- `CAI_MODEL` - Default model - `CAI_MODEL` - Default model
- `CAI_AGENT_TYPE` - Default agent - `CAI_AGENT_TYPE` - Default agent
- `CAI_MAX_TURNS` - Maximum interaction turns - `CAI_MAX_TURNS` - Maximum interaction turns
@ -496,12 +529,14 @@ Display environment variables relevant to CAI.
Execute shell commands directly from the TUI. Execute shell commands directly from the TUI.
**Syntax**: **Syntax**:
``` ```
/shell <command> /shell <command>
$<command> $<command>
``` ```
**Examples**: **Examples**:
```bash ```bash
# List files # List files
/shell ls -la /shell ls -la
@ -514,31 +549,21 @@ $nmap -sV 192.168.1.1
``` ```
**Notes**: **Notes**:
- Commands execute in the system shell - Commands execute in the system shell
- Output is displayed in the terminal - Output is displayed in the terminal
- Use with caution - no sandboxing - Use with caution - no sandboxing
### `/kill` ### Stopping work in the TUI
Terminate the currently executing agent operation. Use **Ctrl+C** to cancel the **current agent turn** or tool call in the focused terminal. Partial responses may be discarded depending on the tool; message history is preserved unless you also `/flush`. To signal an external PID, use **`/shell`** or **`$`** with the usual OS tools.
### `/clear` (scrollback)
Clear **visual** terminal output (scrollback) in the TUI.
**Syntax**: **Syntax**:
```
/kill
```
**Keyboard Shortcut**: `Ctrl+C`
**Notes**:
- Stops agent mid-execution
- Partial responses are discarded
- Agent context is preserved
### `/clear`
Clear the terminal output.
**Syntax**:
``` ```
/clear /clear
``` ```
@ -546,16 +571,21 @@ Clear the terminal output.
**Keyboard Shortcut**: `Ctrl+L` **Keyboard Shortcut**: `Ctrl+L`
**Notes**: **Notes**:
- Clears visual output only - Clears visual output only
- Conversation history is preserved - **Does not** wipe conversation history—use `/flush` / `/clear` in the history sense (see above)
- Cost tracking continues - Cost tracking continues
### `/exit` or `/quit` or `/q`
**Keyboard Shortcut**: `Ctrl+Q` Leave the TUI session cleanly (telemetry flush, session teardown).
**Keyboard Shortcut**: `Ctrl+Q` (depending on build; confirm in-app shortcuts)
**Notes**: **Notes**:
- Unsaved sessions will be lost
- Graceful shutdown of all terminals - Unsaved work in buffers may be lost—`/save` or export logs first if needed
- Shuts down all terminals in the layout
--- ---
@ -568,6 +598,7 @@ Access the command palette for quick command search and execution.
**Keyboard Shortcut**: `Ctrl+P` **Keyboard Shortcut**: `Ctrl+P`
**Features**: **Features**:
- Fuzzy search for commands - Fuzzy search for commands
- Command descriptions - Command descriptions
- Keyboard navigation (arrow keys, Enter) - Keyboard navigation (arrow keys, Enter)
@ -582,7 +613,6 @@ Show or hide the sidebar.
**Alternative**: Click the `[≡]` button in the top bar **Alternative**: Click the `[≡]` button in the top bar
### Clear Input ### Clear Input
Clear the prompt input field. Clear the prompt input field.
@ -590,6 +620,7 @@ Clear the prompt input field.
**Keyboard Shortcut**: `Ctrl+U` **Keyboard Shortcut**: `Ctrl+U`
**Use Cases**: **Use Cases**:
- Parallel agent execution - Parallel agent execution
- Comparing agent responses - Comparing agent responses
- Team-based workflows - Team-based workflows
@ -599,6 +630,7 @@ Clear the prompt input field.
Cancel running operations. Cancel running operations.
**Keyboard Shortcuts**: **Keyboard Shortcuts**:
- `Ctrl+C` - Cancel execution in focused terminal - `Ctrl+C` - Cancel execution in focused terminal
- `Escape` - Cancel all running agents (press twice to exit) - `Escape` - Cancel all running agents (press twice to exit)
@ -606,6 +638,7 @@ Cancel running operations.
## Next Steps ## Next Steps
- [CLI commands reference](../cli/commands_reference.md) — recommended for exact REPL syntax
- [Terminals Management](terminals_management.md) - Advanced multi-terminal workflows - [Terminals Management](terminals_management.md) - Advanced multi-terminal workflows
- [Keyboard Shortcuts](keyboard_shortcuts.md) - Complete keyboard reference - [Keyboard Shortcuts](keyboard_shortcuts.md) - Complete keyboard reference
- [User Interface Guide](user_interface.md) - Visual components and layouts - [User Interface Guide](user_interface.md) - Visual components and layouts
@ -614,5 +647,4 @@ For questions or issues, visit [CAI GitHub Issues](https://github.com/aliasrobot
--- ---
*Last updated: October 2025 | CAI TUI v0.6+* *TUI UX and shortcuts; authoritative slash-command tables: [CLI commands reference](../cli/commands_reference.md).*

2
docs/tui/terminals_management.md
View File

@ -480,8 +480,6 @@ Active ──[Focus Terminal]──> Focused ──[Focus Other]──> Active
**Method 2**: `Escape` twice (all terminals) **Method 2**: `Escape` twice (all terminals)
**Method 3**: `/kill` command
#### Clearing Errors #### Clearing Errors
**Command**: `/clear` (clears visual errors, preserves history) **Command**: `/clear` (clears visual errors, preserves history)

2
mkdocs.yml
View File

@ -30,6 +30,7 @@ nav:
- Welcome: index.md - Welcome: index.md
- Installation: cai_installation.md - Installation: cai_installation.md
- Quickstart: cai_quickstart.md - Quickstart: cai_quickstart.md
- Commands reference: cli/commands_reference.md
# ======================================== # ========================================
# CAI PRO (EXPANDIDO - COMERCIAL) ⭐ # CAI PRO (EXPANDIDO - COMERCIAL) ⭐
@ -119,7 +120,6 @@ nav:
- Tracing & Debugging: tracing.md - Tracing & Debugging: tracing.md
- Context Management: context.md - Context Management: context.md
- Guardrails & Security: guardrails.md - Guardrails & Security: guardrails.md
- LLM Council: council.md
- Environment Variables: environment_variables.md - Environment Variables: environment_variables.md
# ======================================== # ========================================