Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward local text-cleanup helper, but users should be aware it can automatically rewrite longer outgoing prose and overwrite the chosen output file.
Install this only if you want longer prose to be automatically cleaned for AI-writing patterns. Review important, legal, medical, security, business, or confidential text before sending, and use a new output filename when running the script so you do not accidentally replace an existing draft.
The skill instructs the agent to invoke a local script that reads from and writes to files, but the manifest does not declare any tool scope or permissions. That creates a mismatch between advertised and actual capabilities, which can lead to overbroad file access being granted implicitly and makes review, sandboxing, and policy enforcement harder.
The skill says to load automatically for any user-facing prose longer than about 150 words, which is a very broad trigger and may cause it to process sensitive drafts, private communications, or security-relevant content without explicit user intent. Because it rewrites content before sending, over-activation increases the chance of unintended modification, privacy exposure to the script, and bypass of normal review controls.
The script writes transformed content to a user-specified output path, but there is no prompt, warning, or visible disclosure that an existing file at that path may be overwritten. While file output is part of the tool's purpose, the code does not provide any user-facing notice at the point of the write about this data-affecting operation.
No suspicious patterns detected.
