Pa Onboarding

Step-by-step onboarding guide for setting up a new AI Personal Assistant on OpenClaw. Use when: a new PA is being created, someone asks how to set up an agen...

This guide is close to what you'd expect for onboarding an AI PA, but review and tighten before installing or following it: - Confirm required tools and permissions: the SKILL.md assumes presence of the 'gog' CLI, OpenClaw console access, ability to restart the gateway, and permission to edit data/pa-directory.json and create cron jobs. Verify these are available and document them in the registry entry. - Fix the conflicting directives: the guide both requires confirming steps and tells the agent to execute inferred intents without asking. Decide on a strict confirmation policy (prefer explicit owner consent for actions that access accounts or contact people). - Handle credentials securely: do not store API tokens in plaintext files without access controls. Use a secrets manager or at least restrictive file permissions, and document where credentials are stored and how they are rotated. - Get explicit owner consent for contacting third parties and for autonomous actions, and enable auditing/logging of those operations. - If you will let an agent act autonomously, restrict which actions it can perform until onboarding is validated. If you need higher assurance, ask the publisher for a short manifest listing required binaries, exact CLI versions, and recommended secure storage/permission practices. If the publisher cannot provide that, treat the guide as informational only and perform sensitive steps manually.

Type: OpenClaw Skill Name: pa-onboarding Version: 1.1.1 The skill bundle facilitates the onboarding of a personal assistant with broad access to sensitive data, including Gmail, Google Drive, and Calendar. It explicitly instructs the agent to store a monday.com API token in a plaintext file at `~/.credentials/monday-api-token.txt` (SKILL.md), which is a significant security vulnerability. While these high-privilege capabilities and shell-based credential management (using the `gog` utility) are aligned with the stated purpose, the lack of secure secret handling and the request for extensive OAuth scopes represent a high-risk profile.