Recruiter Assistant (Shenzhen)

A professional recruitment workflow assistant. Evaluates resumes against dynamic requirements and AI proficiency, provides critical Pros/Cons analysis, and p...

This skill mostly does what its description says, but don’t install it blindly. Before using: (1) Confirm and declare required binaries (pdftotext) in metadata and ensure they come from trusted packages. (2) Treat all candidate files as sensitive PII — run in a sandbox and limit where output can go. (3) Fix or audit shell calls: the scripts concatenate user-controlled filenames into execSync; sanitize or use spawn with argument arrays to avoid command injection. (4) Clarify external integrations: the README asks agents to send messages to HR and process_incoming accepts a docToken, but no credentials or message-call code are provided—decide how messaging/auth is handled and add required env vars. (5) Test with synthetic resumes first. If you are not able to validate or fix the above, consider this skill suspicious and avoid running it on real candidate data.

Type: OpenClaw Skill Name: recruiter-assistant-sz Version: 1.0.0 The skill bundle contains multiple shell injection vulnerabilities in `scripts/batch_screen.js` and `scripts/process_incoming.js` due to the use of `child_process.execSync` with unsanitized user-controlled input, potentially leading to Remote Code Execution. Additionally, `scripts/generate_questions.js`, `scripts/screen_resume.js`, and `scripts/summarize_interview.js` are vulnerable to path traversal via `fs.readFileSync` with user-provided file paths. While there is no evidence of intentional malice, these critical vulnerabilities could be exploited to compromise the system.