The Agent Skills Directory

[SAFE]: Network security is enforced using curl with mandatory HTTPS, restricted redirection, and response size limits to mitigate resource exhaustion and man-in-the-middle attacks.\n- [COMMAND_EXECUTION]: The skill prevents shell injection by avoiding manual string concatenation for parameters; instead, it utilizes curl's data-encoding capabilities to safely pass user input to the Confluence API.\n- [CREDENTIALS_UNSAFE]: Credential handling is secured through a dedicated environment loader that utilizes an allowlist for environment variables, checks for insecure file permissions on .env files, and rejects values containing control characters.\n- [DATA_EXFILTRATION]: Input validation and URL verification ensure that the skill only interacts with the configured Confluence instance, and no code is present to transmit retrieved data to external endpoints.

confluence-integration