Sendook Email (Restricted)

Read and send emails from an existing Sendook inbox. Use when an AI agent needs to check for new emails, read messages, reply to conversations, or send new emails from a pre-configured inbox. Limited to message operations only — no inbox creation, domain management, or webhook configuration.

This skill appears coherent for its purpose, but before installing: (1) verify the Sendook SDK package (@sendook/node) and its upstream repository are legitimate (check the npm publisher and GitHub repo), (2) create and use a least-privileged API key scoped only to the inbox you want the agent to access, (3) avoid placing the API key in shared or committed files — store it in a secure environment variable store, (4) confirm any local files to attach explicitly with the user before reading them (the SKILL.md also recommends this), and (5) remember that installing the SDK (npm install) will pull third-party code — revoke the API key if you stop using the skill or suspect compromise.

Type: OpenClaw Skill Name: sendook-openclaw Version: 1.1.0 The skill is designed to interact with the Sendook email API for reading and sending emails, which aligns with its stated purpose. Crucially, the `SKILL.md` contains explicit security instructions for the AI agent, warning against reading sensitive local files (e.g., `~/.ssh`, `~/.env`) for attachments and limiting the skill's operational scope. There is no evidence of malicious intent, unauthorized data exfiltration, or prompt injection attempts to subvert the agent's security controls; instead, the instructions enhance security.