Security audit
Security checks for vulnerabilities and agentic risk
This resume skill processes local resume files and writes local analysis/output files, with no evidence of hidden network access, persistence, privilege escalation, or destructive behavior.
Install only if you are comfortable with resume contents being written to local JSON, .docx, and optional backup files. Review and delete intermediate files when finished, and do not expect real Word tracked changes or native Word comments from the current implementation.
This second mismatch finding indicates the declared skill behavior is materially broader than the actual implementation, including absent resume-vs-JD analysis, ATS scoring, keyword generation, and revised .docx output. Such overclaiming can cause dependent systems to skip needed checks or route sensitive user documents into a pipeline that does not deliver the expected security or functional guarantees.
This second mismatch finding indicates the declared skill behavior is materially broader than the actual implementation, including absent resume-vs-JD analysis, ATS scoring, keyword generation, and revised .docx output. Such overclaiming can cause dependent systems to skip needed checks or route sensitive user documents into a pipeline that does not deliver the expected security or functional guarantees.
The skill instructs use of local scripts that read and write files, but it does not declare any explicit tool scope or permissions. This creates an authorization and review gap: an agent may be induced to access or modify user files without clear policy boundaries, which is especially sensitive for resumes and generated documents containing personal data.
The description says the skill provides resume analysis and optimization specifically 'for UK job market.' Under the policy rule, forcing a specific locale without user opt-in is a natural-language policy violation unless the regional restriction is clearly justified, which is not provided here.
The manifest says the skill can generate an improved version of the resume with tracked changes and annotations and supports .docx input/output. In this implementation, the code only analyzes parsed resume JSON plus optional JD text and writes an analysis_report.json; there is no resume rewriting, annotation, change tracking, or .docx handling.
The CLI usage and file operations show the script expects a parsed_resume.json input, an optional jd.txt, and emits a JSON report. No code parses or writes .docx documents, which is a direct mismatch with the manifest's stated file format support.
The module docstring says it will 'Generate optimized resume with Word comments explaining changes,' which implies native Word comment functionality. However, the implementation explicitly notes that native comments are unsupported and instead appends visible inline text markers to paragraphs, so the documentation overstates what the script does.
The add_comment function's docstring states 'Add a comment to a paragraph,' but the code does not create a Word comment object. Instead, it appends a new run containing '[COMMENT: ...]' directly into the paragraph body, which changes visible document content and is materially different from a comment.
This code saves an optimized resume document to the user-specified output path, which is a file write involving sensitive personal information. Although the CLI prints a success message after saving, there is no prior warning, confirmation, or explanatory comment at the point of execution informing the user that resume data will be written to disk.
When --backup is provided, the script creates a duplicate of the original resume file, which increases the number of stored copies of sensitive personal data. The code reports success afterward, but it does not provide any advance disclosure or caution that a full backup of the resume will be created on disk.
The script writes parsed resume content, including potentially sensitive personal data, to a JSON file on disk by default without any consent prompt, minimization, or retention controls. In the context of a resume analyzer, this is especially risky because resumes commonly contain names, contact details, employment history, and other personal information that may be exposed through shared workspaces, backups, logs, or leftover temporary files.
The document is explicitly framed as 'UK Professional Resume Templates' and repeatedly mandates UK-specific formatting and conventions such as postcode usage, UK grading terms, and UK resume norms. Under the stated policy, forcing a locale-specific format without offering user choice or opt-in is a natural-language policy violation unless clearly documented as a justified region-specific tool.
No suspicious patterns detected.
