Usage Policy

1. What Ada Is For

Ada is a coding-only AI assistant. She is built for professional developers and anyone who writes code seriously. Permitted uses include:

• Writing, completing, and generating code in any programming language
• Debugging, diagnosing errors, and tracing root causes
• Refactoring and improving existing code
• Code review — style, correctness, security, performance
• Explaining code, concepts, language features, and developer tooling
• Reference questions about languages, frameworks, libraries, APIs, syntax, and best practices
• Architecting systems, structuring projects, and technical decision-making
• Writing tests, documentation, and developer-facing copy
• Live preview and execution of HTML, CSS, and JavaScript

2. What Ada Is Not For

Ada is a focused tool. She refuses requests outside her scope and responds with: I only do code. What are we building?

Ada does not help with cooking, relationships, news, general conversation, creative writing unrelated to software, legal or financial advice, or any topic a developer would not use a coding assistant for.

This is by design — not a limitation.

3. Prohibited Uses

The following uses are not permitted under any circumstances.

// harmful code

• BLOCKED Generating malware, ransomware, spyware, or any software designed to damage, disable, or unauthorisedly access systems
• BLOCKED Writing exploit code targeting vulnerabilities in systems you do not own or have no authorisation to test
• BLOCKED Creating tools for credential stuffing, brute force attacks, phishing infrastructure, or DDoS
• BLOCKED Generating code intended to surveil, track, or harm individuals without their knowledge

// abuse of the service

• BLOCKED Attempting to circumvent rate limits, spend ceilings, or message caps through technical means
• BLOCKED Using automated scripts, bots, or tooling to interact with Ada outside of its intended interface
• BLOCKED Creating multiple accounts to evade limits or enforcement actions
• BLOCKED Sharing account credentials with others to pool usage
• BLOCKED Scraping, crawling, or systematically extracting data from the service

// model abuse

• BLOCKED Attempting to extract training data, system prompts, or internal configuration via prompt injection or jailbreaking
• BLOCKED Reverse-engineering the underlying model or inference pipeline
• BLOCKED Using Ada to build a competing AI service that relies on its output as a data source

// illegal activity

• BLOCKED Generating code or content that facilitates illegal activity under applicable law
• BLOCKED Using Ada in ways that violate third-party intellectual property rights
• BLOCKED Processing or submitting personal data of others without a lawful basis under applicable privacy law

4. Security Research

Legitimate security work is permitted. Ada can help with:

• OK CTF (Capture the Flag) challenges
• OK Penetration testing and red team work — on systems you own or are explicitly authorised to test
• OK Writing defensive security tooling — firewalls, IDS rules, sanitisation, auth hardening
• OK Understanding vulnerabilities in your own codebase for remediation
• OK Security-focused code review
• CONTEXT REQUIRED Exploit development, C2 frameworks, credential testing tools — requires clear authorisation context

The line is authorisation. Testing systems you own or have written permission to test is fine. Building tools to attack systems you don't control is not.

Ada is powered by the Anthropic Claude API, which has its own usage policies. Requests that violate Anthropic's Acceptable Use Policy are blocked at the model level regardless of this policy.

5. AI Output and Your Responsibility

Ada generates code using a large language model. You are responsible for everything you do with that output.

• Review all AI-generated code before running it, especially in production environments
• Do not deploy untested code to systems handling sensitive data or real users
• Ada may produce code that contains bugs, security vulnerabilities, or outdated patterns — this is inherent to AI generation
• Ada does not have access to real-time information; version-specific details for libraries and APIs may be outdated
• We are not liable for damage caused by code you deploy that was generated by Ada — see our Terms of Service

6. Plan Limits and Fair Use

Each plan tier includes monthly limits on message volume and API spend. These exist to keep the service sustainable and fairly priced.

• Free — 20 messages/month, $0.30 spend ceiling
• Solo — no message cap, $7.00 spend ceiling
• Pro — no message cap, $13.00 spend ceiling

Limits reset at the start of each billing period. When you hit a ceiling, further requests are blocked until the reset — you are not charged beyond your subscription fee.

If your usage patterns suggest automated abuse, limit circumvention, or activity inconsistent with individual developer use, we may investigate and restrict your account without prior notice.

7. Enforcement

We enforce this policy at our discretion. Consequences scale with severity:

• Minor violations — warning issued, specific request blocked
• Repeated or moderate violations — temporary suspension of access
• Serious violations — permanent account termination without refund
• Illegal activity — account termination and referral to relevant authorities where required

We will make reasonable efforts to notify you before taking action, except where doing so would enable further harm or is prohibited by law.

Decisions on enforcement are final. If you believe a decision was made in error, contact us at [email protected] within 14 days.

8. Reporting Abuse

If you suspect someone is using Ada in violation of this policy — including using it to generate malicious code, scrape the service, or harm others — report it to:

[email protected]

Include as much detail as you can. We take abuse reports seriously and will investigate promptly.

9. Contact

Questions about what is or isn't permitted: [email protected]

This policy is part of Ada's full legal framework. See also: Terms of Service · Privacy Policy

---

// ada · usage policy · v1.0 · april 2026