Frequently Asked Questions

This page answers the most common questions about RTILA X. Questions are organized by category.

General

1. What is RTILA X?

RTILA X is a local-first, AI-powered web automation platform that runs entirely on your computer. It lets you record, build, and run browser automations — clicking buttons, filling forms, scraping data, crawling websites — without writing code. You describe what you want in plain English, and the AI builds the automation for you.

2. Is RTILA X free?

Yes. RTILA X offers a free community license that includes:

1 device activation
1 active automation project
Unlimited runs for that 1 project
Access to the AI Assistant

Paid plans start at $9/month or $99 lifetime and include unlimited projects, all features, and priority support.

3. Does RTILA X work offline?

Yes, mostly. RTILA X runs entirely on your computer. However:

Initial setup requires an internet connection (database initialization, license activation).
Web scraping requires an internet connection to access target websites.
Cloud AI providers (OpenRouter) require an internet connection.
Local AI models require an internet connection only for the initial model download. After download, no internet is needed.
CAPTCHA solving requires an internet connection (2Captcha service).

4. What platforms are supported?

Platform	Versions
Windows	Windows 10 (version 1809+), Windows 11, Windows Server 2019+
macOS	macOS 11 (Big Sur)+, Apple Silicon and Intel
Linux	Ubuntu 22.04+ (full Desktop GUI required, GNOME recommended)

5. What browsers are supported?

RTILA X supports Microsoft Edge, Brave Browser, and Google Chrome. Microsoft Edge is recommended because it’s pre-installed on Windows and has the best compatibility with RTILA X’s stealth features.

6. Where is my data stored?

All data is stored locally in ~/.rtila-x/ on your computer. This includes:

Projects, sessions, logs, notifications
Chat history
Browser profiles
AI models
Output files
Backups

No data is sent to external servers unless you explicitly configure triggers to send data.

AI Features

1. What AI providers does RTILA X support?

RTILA X supports two AI providers:

OpenRouter — Access to hundreds of cloud AI models through a single API key. Requires an internet connection.
Local LLM — Run any GGUF model from Hugging Face locally using llama.cpp. No internet needed after download.

2. Can I use my own AI model?

Yes. The Local LLM provider supports:

Any GGUF model from Hugging Face in the user/repo/file.gguf format.
Any OpenAI-compatible API endpoint (Ollama, LM Studio, vLLM, text-generation-webui, etc.).

3. What happened to the finetuned models?

The finetuned models (RTILA Lite 1.5 Assistant) were removed in v8.3.2. They were replaced by support for any GGUF model from Hugging Face. See Release Notes for details.

4. What happened to RAG?

RAG (Retrieval-Augmented Generation) was removed in v8.3.2. It was replaced by full-context injection, where the complete system prompt and command reference are injected into every AI conversation. This is more reliable, faster, and uses fewer resources. See Release Notes for details.

5. Can the AI build automations for me?

Yes. The AI Assistant can build complete automations from plain English descriptions. Just describe what you want to automate, and the AI will:

Analyze the target page.
Generate the project configuration (commands, datasets, settings).
Create the project for you to run.

Yes. When the AI encounters a login wall, it prompts you for credentials. When it encounters a CAPTCHA, it can solve it using 2Captcha (if configured). The AI handles these automatically as part of the automation building process.

Automation

1. How do I create an automation?

There are four methods:

AI Assistant — Describe what you want in plain English.
Record — Record your browser actions as commands.
Manual — Build commands one by one in the Visual Builder.
Templates — Import a pre-built template.

2. How do I run an automation?

Go to Automation Projects.
Select your project.
Click Run in the ribbon toolbar.
The Live Monitor opens showing real-time progress.

3. Can I schedule automations?

Yes. Use the Task Scheduler to schedule automations to run at specified intervals (every N minutes, hours, daily, or weekly). See Task Scheduler for details.

4. What happens if an automation fails?

If an automation fails:

The session status is set to “failed”.
The error message is logged.
Triggers still fire (unless the run was stopped or killed).
You can check the Execution Logs for details.
You can use checkpoints to resume from where it left off.

Data

1. What formats can I export data in?

RTILA X supports exporting data in these formats:

CSV — Opens in Excel, Google Sheets
JSON — Standard data format
XLSX — Excel spreadsheet
Parquet — Columnar storage for analytics
XML — XML format
JSONL — JSON Lines format

2. Where is my scraped data stored?

Scraped data is stored in the PocketBase database at ~/.rtila-x/database/data/data.db. You can view it on the Scraped Data page. Exported files are saved to ~/.rtila-x/output/.

3. Can I send data to a database?

Yes. RTILA X supports database triggers for PostgreSQL, MySQL, and MongoDB. You can also use triggers to send data to cloud storage, APIs, email, and 30+ other destinations. See Triggers for details.

Stealth

1. How do I avoid getting blocked?

RTILA X includes several stealth features:

Proxy Management — Route traffic through proxy servers to mask your IP.
CAPTCHA Solving — Automatically solve CAPTCHAs using 2Captcha.
Humanoid Behavior — Simulate human-like mouse movements and typing.
Browser Fingerprint Alignment — Match timezone/locale to proxy location.
Browser Profiles — Isolated browser environments with unique fingerprints.

2. What is humanoid behavior?

Humanoid behavior simulates human-like interactions: curved mouse movements, variable typing speeds, slight positional randomness, and natural pauses. It’s enabled by default and is the most effective anti-detection feature.

3. Do I need proxies?

Proxies are recommended for high-volume scraping or websites with aggressive anti-bot protection. For light scraping of unprotected websites, proxies may not be necessary. See Proxy Management for details.

Standalone Bots

1. What is a Standalone Bot?

A Standalone Bot is a packaged automation that runs independently without RTILA X installed. You can distribute it to clients, deploy it on servers, or sell it as a product.

2. How do I create a Standalone Bot?

Go to Standalone Bots in the sidebar.
Click New Bot.
Configure the bot (name, source project, visibility, password).
Click Save & Upload.
Click Download Archive to get the executable.

3. Can I password-protect a bot?

Yes. When creating a bot, set the visibility to “Password Protected” and set a password. Users will need to enter the password to run the bot. You can change the password at any time to revoke access.

Local LLM

1. What is the Local LLM?

The Local LLM runs AI models locally on your computer using llama.cpp. No internet connection is needed after the model is downloaded. Your data never leaves your machine.

2. What models are supported?

Any GGUF model from Hugging Face is supported. Recommended models include:

bartowski/Qwen2.5-7B-Instruct-GGUF/Qwen2.5-7B-Instruct-Q4_K_M.gguf
bartowski/Meta-Llama-3.1-8B-Instruct-GGUF/Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf
bartowski/Mistral-7B-Instruct-v0.3-GGUF/Mistral-7B-Instruct-v0.3-Q4_K_M.gguf

3. How much RAM do I need?

Model Size	Minimum RAM	Recommended RAM
7B parameters	6 GB	8 GB
13B parameters	10 GB	16 GB
30B+ parameters	20 GB	32 GB+

4. What is KV cache?

KV (Key-Value) cache stores the attention state of processed tokens. By tuning the cache, you can reduce memory usage and improve response speed. RTILA X provides full control over KV cache settings including quantization type, Flash Attention, and parallel slots.

5. What is Flash Attention?

Flash Attention is a memory-efficient attention algorithm that reduces peak memory during generation. It’s required for quantized KV cache on GPU backends. Enable it if your GPU supports it.

6. My local LLM won’t start. What do I do?

See Troubleshooting for detailed troubleshooting steps. Common fixes:

Reduce context size to 8K or 4K.
Set GPU Layers to 0 (CPU only).
Re-download the model.
Switch to OpenRouter if local LLM continues to fail.

Trigger Chains

1. What is a Trigger Chain?

A Trigger Chain is a reusable multi-step post-processing pipeline. It runs after a project run completes and can include conditional logic (IF/ELSE), parallel execution, and scheduling. See Trigger Chains Overview for details.

2. How is a Trigger Chain different from project triggers?

Feature	Project Triggers	Trigger Chains
Scope	Per-project	Standalone, reusable
Conditional logic	No	IF/ELSE blocks
Parallel execution	No	Parallel blocks
Scheduling	No	Yes
Webhook triggers	No	Yes
Run history	Limited	Detailed tracking

3. Can I schedule a Trigger Chain?

Yes. Trigger Chains can be scheduled to run at specified intervals (every N minutes, hours, daily, or weekly). See Trigger Chain Scheduling for details.

MCP Server

1. What is the RTILA MCP server?

The RTILA engine binary (rtila-engine) doubles as a Model Context Protocol (MCP) server when started with the --mcp flag. It exposes 17 tools that let external AI clients (Claude Code, Cursor, VS Code, Windsurf, etc.) analyze web pages, run automation projects, and read scraped data — all through the standard MCP protocol.

2. Do I need to install anything extra for the MCP server?

No. The MCP server is built into the same rtila-engine binary that runs your automations. There is no second binary, no separate download, and no additional installation. You just need the RTILA X application installed and an MCP-compatible client.

3. Which MCP clients are supported?

Any client that supports stdio-based MCP servers works. Tested clients include Claude Code, Claude Desktop, Cursor, VS Code with GitHub Copilot, Windsurf, Continue.dev, and Cline. See MCP Client Setup for configuration instructions.

4. Does the MCP server replace the AI Assistant?

No. The AI Assistant (built into the RTILA X desktop app) and the engine MCP server are separate interfaces. The AI Assistant is for interactive automation building within RTILA X. The MCP server is for external AI tools to drive RTILA programmatically. See MCP Workflows — MCP Server vs. AI Assistant Tools for the comparison.

5. Why do browser tools return a `run_id` instead of results?

Browser tools (page analysis, project runs) take minutes to complete, and most MCP clients kill tool calls at about 60 seconds. The engine returns a run_id immediately and you poll rtila_get_run_status until the run finishes. This is the async contract. See MCP Workflows — The Async Contract for details.

6. Can I use the MCP server without the RTILA X desktop app open?

Yes. The desktop app does not need to be open. However, the local PocketBase database must be running and reachable at the URL you pass as --rest-api. Only the two analysis tools (rtila_analyze_page, rtila_analyze_api_endpoint) work without a database connection.

7. Is the MCP server secure?

The server is designed for loopback (local) use. It exposes your local database and can drive a real browser with your proxy and CAPTCHA settings. Treat the RTILA_TOKEN like a password and never expose the database port externally. See MCP Workflows — Security for details.

Troubleshooting

1. My browser won’t launch. What do I do?

See Troubleshooting — Browser Won’t Launch for detailed troubleshooting steps. Common fixes:

Ensure a compatible browser is installed (Edge, Brave, Chrome).
Check the browser path in Preferences.
Check antivirus settings.
Try a different browser.

2. My automation fails with “Element not found”. What do I do?

The selector doesn’t match any element on the page. Common fixes:

Use the Element Picker to regenerate the selector.
Check for page structure changes.
Check for iframes.
Add a wait_for_selector command before the failing command.

3. My automation is too slow. What do I do?

Common fixes:

Reduce humanoid sensitivity to Low.
Enable headless mode.
Reduce the number of concurrent workers.
Reduce the navigation timeout.
Use a smaller AI model or reduce context size.

4. How do I reset RTILA X?

See Troubleshooting — Resetting the Application for details. Options:

Soft Reset — Delete browser profiles only.
Full Reset — Delete the entire ~/.rtila-x/ directory.
Reset License Only — Reset license data only.