Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed research-and-documentation workflow for OysterWorkflow demo planning, with no hidden code or malicious behavior found.
Before installing, confirm you want an agent to use your signed-in Chrome session, access ChatGPT or similar research sources, browse public YC/video pages, and create or edit a Google Doc. Verify the exact target document before allowing edits, especially if a document named “demo 计划 1” already exists.
The activation guidance is broad enough to overlap with ordinary research, marketing, and planning work, which can cause the agent to invoke this skill in situations the user did not specifically intend. Because the skill performs downstream actions like browsing and editing documents, ambiguous triggering increases the chance of unintended external actions and document modifications.
The skill explicitly states there are no exclusions, leaving no guardrails for adjacent tasks where this workflow may be inappropriate. In practice, that ambiguity can cause over-application of a skill that opens sites and updates an existing Google Doc, creating a real risk of unintended modifications or unnecessary data exposure.
The skill directs the agent to use and update a named Google Doc but does not require an explicit user confirmation immediately before modifying that existing document. This is dangerous because it can lead to silent overwrites, contamination of prior notes, or edits to the wrong document if multiple similarly named files exist.
The guidance suggests a specific English CTA: 'Send us a workflow your team still does by hand. We’ll turn it into an AI skill.' Elsewhere the document mixes English and Chinese, but it does not state that output language should follow user preference or that English is required for a justified business reason.
No suspicious patterns detected.
