Terms of Service
Last updated: June 14, 2026
AgentVigil can send keystrokes to terminal processes running on your Mac. Always understand what you are approving before tapping the approve button. AI coding agents can execute destructive commands.
Acceptance
By installing or using AgentVigil (the Mac companion software and the Android application, collectively "the Software"), you agree to these terms. If you do not agree, do not use the Software.
What AgentVigil Is
AgentVigil is a developer tool that:
- Monitors AI coding agent sessions (Claude Code, Codex CLI, Amp, and others) running on your Mac
- Sends push notifications to your Android phone when sessions require your attention
- Allows you to respond to permission prompts from your phone
- Tracks token usage and cost data from session transcripts on your local filesystem
AgentVigil is not affiliated with, endorsed by, or operated in partnership with Anthropic (Claude Code), OpenAI (Codex), or any other AI company whose products it monitors.
License
Open source under the MIT License. You are free to use, copy, modify, merge, publish, distribute, sublicense, and sell copies of the software subject to the conditions in the LICENSE file in the GitHub repository.
Proprietary software. You are granted a personal, non-transferable, non-exclusive license to install and use the app on Android devices you own or control. You may not reverse engineer, decompile, or redistribute the Android app.
What You Are Responsible For
You are solely responsible for:
- The AI agent sessions running on your Mac and the commands they execute
- Decisions to approve or deny permission prompts, whether made in the terminal or via the AgentVigil app
- Any costs incurred with Anthropic, OpenAI, or other AI providers as a result of your agent sessions
- Securing your paired device and ensuring unauthorized persons do not have access to your phone
- Keeping the Mac companion software up to date
AgentVigil relays your approve and deny decisions to your terminal. We are not responsible for the consequences of those decisions.
No Warranty
The Software is provided "as is" without warranty of any kind, express or implied, including but not limited to warranties of merchantability, fitness for a particular purpose, and non-infringement.
We do not warrant that:
- The Software will be uninterrupted or error-free
- Permission notifications will always be delivered in time
- Approve and deny commands will always reach your terminal successfully
- Token usage and cost calculations will be exactly accurate
- The Software will be compatible with future versions of Claude Code, Codex, or other AI tools
AI coding agents can execute destructive commands. Always understand what you are approving before tapping the approve button.
Limitation of Liability
To the maximum extent permitted by applicable law, AgentVigil and its developers shall not be liable for any indirect, incidental, special, consequential, or punitive damages, including but not limited to:
- Data loss or corruption caused by commands approved via the app
- API costs incurred during AI agent sessions
- Loss of code, work, or intellectual property
- Missed permission prompts due to notification delivery failures
- Any other damages arising from use or inability to use the Software
Our total liability to you for any claim arising from use of the Software shall not exceed the amount you paid for the Software in the twelve months preceding the claim. Since AgentVigil is currently free, that amount is zero.
Third-Party Services
AgentVigil uses the following third-party services:
| Service | Purpose | Terms |
|---|---|---|
| Cloudflare Tunnel | Encrypted connection between Mac and phone | cloudflare.com/terms |
| ntfy.sh | Push notification delivery | ntfy.sh/terms |
| Firebase (FCM) | Android push notifications | firebase.google.com/terms |
We are not responsible for the availability, reliability, or terms of these third-party services. If these services change their terms or become unavailable, AgentVigil may stop functioning.
Acceptable Use
You agree not to use AgentVigil to:
- Monitor AI agent sessions on machines you do not own or have explicit permission to access
- Intercept or monitor another person's AI agent sessions without their knowledge and consent
- Circumvent security controls on any system
- Violate any applicable law or regulation
The Software Accesses Your Terminal
You understand and accept that:
- The AgentVigil Mac companion registers hooks with Claude Code and other AI tools installed on your Mac
- These hooks give the companion visibility into tool calls, permission prompts, and session lifecycle events
- The companion reads JSONL transcript files written by Claude Code to your local filesystem
- The companion can send keystrokes to terminal processes for approve and deny functionality
- The Mac companion source code is fully open source so you can audit exactly what it does
If you do not accept these conditions, do not install the Mac companion. The source code is available on GitHub for review before installation.
Modifications to the Software and Terms
We reserve the right to modify the Software and these Terms at any time. Material changes to these Terms will be announced via the GitHub repository. Continued use of the Software after changes constitutes acceptance.
We reserve the right to discontinue the Software at any time without notice.
Governing Law
These Terms are governed by the laws of the jurisdiction in which the developer is located, without regard to conflict of law principles. Any disputes shall be resolved in the courts of that jurisdiction.
Severability
If any provision of these Terms is found to be unenforceable, the remaining provisions will continue in full force and effect.
Contact
For questions about these Terms, open an issue on GitHub or send us an email.
GitHub Issues: github.com/mdusaama0/agentvigil-site/issues
Email: stacktreelabs@gmail.com
Entire Agreement
These Terms, together with the Privacy Policy, constitute the entire agreement between you and AgentVigil regarding the Software and supersede any prior agreements.
AgentVigil is an independent developer tool. It is not affiliated with Anthropic, OpenAI, or any AI company whose products it monitors. Claude Code, Codex, and Amp are trademarks of their respective owners.