Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward tweet-idea generator that reads user-provided material or URLs and saves a local markdown output file.
Before installing, understand that URLs you provide may be fetched over the network and the output may be fewer than 60 ideas if the source material does not fit every category. Avoid giving private drafts, internal URLs, or sensitive material unless you intend the agent to process them.
The skill instructs the agent to fetch arbitrary user-provided URLs without any safety notice, validation guidance, or restriction on destinations. That can expose user-supplied links and agent network activity to third parties, and in some environments may enable access to internal resources or sensitive metadata endpoints if URL fetching is not sandboxed.
The manifest description at L003 and workflow at L012/L047 present the skill as generating 60 tweet ideas. However, L151-L153, L197-L198, and L293 allow skipping categories when content does not fit, which can reduce the output below 60 and conflicts with the stated promise.
At L076 the skill says to generate exactly 10 statements in each of 5 categories, implying a fixed total and structure. But L150-L153 and L197-L198 explicitly allow skipping categories and redistributing effort, which contradicts the earlier instruction rather than merely adding detail.
No suspicious patterns detected.
