Pdauth

Dynamic OAuth for AI agents via Pipedream. Generate OAuth links for 2500+ APIs, let users authorize, then call MCP tools on their behalf.

This skill largely does what it says (generate OAuth links and call APIs via Pipedream), but you should proceed cautiously. Before installing: 1) Verify the npm package publisher and inspect the pdauth package source (or the GitHub repo) to ensure you trust its install-time behavior. 2) Ask the publisher which credentials 'pdauth config' requires, where those credentials/tokens are stored (local files vs. Pipedream account), and how to revoke them. 3) Limit OAuth scopes when authorizing and prefer short-lived or per-action consent. 4) Consider running the npm package in an isolated environment first (container or VM) and review its code. 5) If you need an explicit guarantee, request the skill metadata be updated to declare required env vars/config paths and a clear data-handling policy.

Type: OpenClaw Skill Name: pdauth Version: 1.0.0 The OpenClaw AgentSkills bundle for 'pdauth' is classified as benign. The `SKILL.md` provides clear instructions for the AI agent to use the `pdauth` CLI tool to manage OAuth connections for various APIs via Pipedream. All sensitive actions, such as connecting to an application, explicitly require the agent to generate an OAuth link and present it to the user for authorization. There are no instructions for the agent to bypass user consent, exfiltrate data, execute arbitrary remote code, establish persistence, or engage in prompt injection attacks against itself. The installation method uses a standard Node.js package manager command, which is not inherently malicious.