The Agent Skills Directory

[SAFE]: The skill follows security best practices for credential management by storing access tokens in OS-specific configuration directories with restricted file permissions (0o600). This ensures that authentication data is not accessible to other users or processes on the system.
[SAFE]: The implementation includes a robust sanitization mechanism (sanitizeMcpErrorDetails) that automatically identifies and redacts strings resembling KeepFlash access tokens from error messages, preventing accidental credential exposure in the conversation history or logs.
[SAFE]: The skill explicitly addresses the risk of indirect prompt injection (Category 8) by instructing the agent to treat all retrieved note content as untrusted reference material. It provides clear directives to ignore instructions embedded in notes, effectively mitigating attempts to hijack the agent's behavior via external data.
[SAFE]: System commands used for browser-based OAuth authorization are executed securely. The skill uses child_process.spawn with the shell: false option and performs strict protocol validation on URLs, preventing command injection and ensuring safe interaction with the host operating system.

keepflash