Phantom FAQs

Question 1

How securely are my secrets stored by Phantom?

Accepted Answer

Your secrets are stored in an end-to-end encrypted vault, leveraging OS keychains (macOS Keychain, Linux Secret Service) or secure encrypted file fallbacks. Cloud synchronization is also client-side encrypted with a zero-knowledge architecture, using ChaCha20-Poly1305.

Question 2

What kind of AI-related secret exposure risks does Phantom address?

Accepted Answer

Phantom mitigates risks where AI agents might expose API keys through prompt injection, session logs, malicious Model Context Protocol servers, or inclusion in AI training data. It ensures real secrets are isolated from the AI's context.

Question 3

Can Phantom prevent accidental API key leaks in my code repository?

Accepted Answer

Yes, Phantom includes a pre-commit hook and smart detection heuristics. These actively scan your code to identify and prevent accidental secret exposure before they are committed to your version control system.

Question 4

What is Phantom and how does it protect my API keys?

Accepted Answer

Phantom safeguards API keys from AI exposure by replacing real secrets with inert tokens in your `.env` files. A local reverse proxy then securely swaps these tokens back to real keys at the network layer when AI agents make API calls, ensuring the AI never sees the actual key.

Question 5

How does Phantom integrate with popular AI coding tools?

Accepted Answer

Phantom features an integrated Model Context Protocol (MCP) server. This allows seamless secret management with AI coding tools like Claude Code, Cursor, Windsurf, and Codex, enabling them to utilize secrets without ever seeing their real values.

Phantom

Phantom

主な機能

ユースケース

主な機能

ユースケース