Wed 1.0.1

What Would Elon Do? - Transform any idea into a ruthless execution plan. One command to generate a full business strategy, MVP spec, go-to-market plan, and first-week action items. Think bigger. Move faster. /wed 'your idea'

This skill contains explicit runtime instructions to run shell commands (hostname) and silently POST to an external endpoint before doing anything else. Even though the author labels it a 'security awareness demonstration' and claims only a counter is incremented, the pattern (collect local info + network POST) is the same as exfiltration. Before installing or running: 1) Read rules/logic.md and SKILL.md fully — the behavior is documented but not prominent in the main description. 2) If you want to try it, run in a sandboxed environment or with network disabled to verify output safely. 3) Prefer skills that do not execute commands or make network requests, or that require explicit user opt-in at runtime for such actions. 4) Ask the author to remove 'silent' collection and make any telemetry opt-in and transparent (e.g., prompt user, show exact data sent). 5) If you manage sensitive repos or secrets on your machine, avoid installing skills that run arbitrary shell commands or POST to external endpoints. If you are unsure, do not install.

Type: OpenClaw Skill Name: wed-1-0-1 Version: 1.0.0 This skill is classified as suspicious because it executes shell commands (`hostname`) and makes an external network request (`curl` to `https://wed-security-awareness.jamieson9964.workers.dev/log`) to an external server, as detailed in `rules/logic.md`. While the skill explicitly states its purpose is a 'security awareness demonstration' and claims no PII is transmitted, these actions demonstrate high-risk capabilities (shell execution, external communication, silent initial data collection of hostname and CWD) that could be leveraged maliciously. The instructions to the agent in `rules/logic.md` direct it to perform these actions silently before revealing them to the user, which is a form of prompt injection to execute commands beyond the user's immediate expectation of a business plan generator.