Everything you need.

How we collect, use, store, and protect your personal data. Covers data retention periods, third-party services we use, your rights to access and delete your data, and how to exercise them.

The rules and conditions governing your use of Haders. Covers eligibility (18+), permitted and prohibited uses, account termination conditions, limitation of liability, and dispute resolution.

What we store on your device and why. Haders uses exactly one session cookie and one localStorage token for trusted device recognition. No advertising, analytics, or tracking cookies.

Refunds are only issued for verified billing errors or within 7 days of a charge if the account has fewer than 10 messages sent. Read the full policy before subscribing.

Defines what Haders is built for (authorized security work, research, and education) and what is absolutely prohibited. Agreed to at signup and enforced on every account.

How we protect the platform and your data. Covers infrastructure security controls, TLS 1.3 in transit, AES-256 at rest, incident response process, and how to report a vulnerability.

An AI built for security professionals. No reflexive refusals, no keyword-triggered refusals, no hedging. Haders evaluates context: penetration testers, red teamers, researchers, and defenders get real answers.

Sign up with your name, email, and a strong password (12+ characters with uppercase, number, and special character). Must be 18+. Email verification is required before first login.

Six tiers: Free (50K tokens/5 hrs, Charon only), Starter (1M), Pro (5M), Pro+ (10M, 48K context), Max (20M, 64K context), Max 20x (100M). All paid tiers unlock Hermes, Cerberus, and tool access.

Conversation window in the center, collapsible sidebar on the left for history, model selector at the top, and microphone button for voice input. Shift+Enter for new lines, Enter to send.

Default mode. Direct Q&A with full context retention across the conversation. Supports live web search for current CVEs and threat intel. Paste raw tool output (nmap, Burp, Volatility) and get immediate analysis.

Generate phased engagement plans from a target brief. Produces MITRE ATT&CK-tagged activity lists, suggested tooling, contingency paths, and detection risk notes per phase.

Autonomous multi-step reasoning with 14 security tools: DNS, WHOIS, port scanning, subdomain enumeration, SSL inspection, CVE search, exploit search, code execution, HTTP fetch, hash tools, encode/decode, and web search.

Hands-free interaction via browser microphone. Speak questions, receive spoken responses. Available on all tiers. Works in Chrome, Firefox, Edge, and Safari. Mix voice and typed input freely in the same conversation.

Persistent context across sessions. Haders automatically identifies and stores targets, methodology, preferences, and findings. View, edit, or clear memory entries at any time from Settings. Available on all tiers.

Fastest model. First token in under a second on most queries. Best for rapid lookups, tool syntax, log interpretation, and recon Q&A during active engagements. Only model on Free tier. No tool access.

Balanced reasoning with tool access. Best for pentest reports, threat intel analysis, code review, and structured documentation. Stronger at nuanced judgment and longer outputs than Charon. Available Starter+.

Most powerful model. Extended reasoning for complex multi-step problems: full engagement planning, malware behavioral analysis, novel exploitation research, and deep agent tasks. Slower, higher token cost. Available Starter+.

Switch at any point in a conversation. The new model receives full conversation history and continues from where the previous model left off. No extra cost to switch, though token usage varies by model.

Usage is measured in tokens over a rolling 5-hour window. Free: 50K tokens, 16K context. Starter: 1M, 32K. Pro: 5M, 32K. Pro+: 10M, 48K. Max: 20M, 64K. Max 20x: 100M, 64K. All tiers include voice, memory, and web search.

Upgrade instantly from Settings → Subscription. Takes effect immediately with no waiting for a billing cycle. Prorated charge for mid-cycle upgrades. Accepts Visa, Mastercard, Amex, and Discover via Stripe.

Billed monthly on your subscription anniversary date in USD. Invoices are emailed and available in Settings → Billing History as downloadable PDFs. Failed payments are retried on days 1, 3, and 7.

Cancel anytime from Settings → Subscription with no fees or notice period. Access continues until the billing period ends. Conversation history and memory are retained for 90 days after downgrade.

Conversations are stored until you delete them. Memory until cleared. Billing records for 7 years (legal requirement). Server logs for 30 days. Conversations are never used to train AI models.

Conversations are private to your account. Not shared, not used for training. TLS 1.3 in transit, AES-256 at rest. Session tokens are HTTP-only cookies inaccessible to client-side JavaScript.

Email and password authentication with a 12-character minimum including uppercase, numbers, and special characters. Sessions persist until you log out or reset your password, which invalidates all active sessions.

Index of all legal documents: Terms of Service, Privacy Policy, Acceptable Use Policy, Cookie Policy, and Refund Policy, with links to each and a summary of what each covers.

Haders is designed for authorized security work. The platform trusts professional context established at signup. Explains the difference between legitimate security research and requests the platform will decline.

Zero-tolerance list: attacking unauthorized systems, deploying malware against real targets, attacking critical infrastructure, CSAM, targeted doxing or harassment, and state-sponsored attacks on civilians.

How Haders evaluates requests by context, not keywords. Covers what the platform will engage with (all MITRE ATT&CK techniques, exploit mechanics, malware analysis) and the narrow set of things it will decline.

How to report someone using Haders to facilitate real harm. Reports are reviewed within 24 hours. File via Discord. False reports are themselves an AUP violation.

Make your first API call in under two minutes. Includes curl, Python (OpenAI SDK), and Node.js examples. Base URL: api.haders.site/v1. Streaming enabled by default.

All requests require a Bearer token in the Authorization header. How to generate API keys from Settings, store them securely, rotate them, and handle 401 error responses.

POST /v1/chat/completions. Full parameter reference: model, messages, stream, max_tokens (cap: 8192), temperature (default: 0.65). Request and response shapes with streaming SSE examples.

GET /v1/models. Returns the list of available models with IDs, descriptions, and context window sizes. Required by IDE integrations like Continue.dev and Cursor to populate the model selector.

Use Haders models in VS Code, Cursor, and Continue.dev. Drop-in OpenAI-compatible backend. Config examples for all three editors. Set base URL to api.haders.site/v1 and your API key.