Recruiter Assistant

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

This skill appears to do what its name says, but proceed cautiously. Key points to check before installing or running: 1) pdftotext is required by the scripts but not declared — install and test it separately. 2) The scripts print entire resume contents to stdout and instruct the agent to create and publish Feishu documents and to post summaries to HR; this will transmit candidate PII to external services. Confirm where Feishu docs are stored, who can access public links, and what authentication/scopes the feishu_doc and message tools use. 3) There are no declared credentials or env vars for Feishu; verify how authentication is handled by your agent runtime and limit permissions/scopes. 4) If you will process real candidate data, run the skill in an isolated environment (or with sanitized/test resumes) until you confirm behavior and storage/privacy controls. 5) Consider asking the publisher to: declare pdftotext as a required binary, document required tool permissions, avoid printing raw resumes to logs, and add explicit guidance or opt-in for publishing public links. If any of these are unacceptable for your privacy/security posture, do not use the skill with real candidate data.

Type: OpenClaw Skill Name: recruiter-assistant Version: 1.4.2 This skill bundle is highly suspicious due to multiple severe vulnerabilities. Several scripts (`scripts/batch_screen.js`, `scripts/process_incoming.js`) use `child_process.execSync` with user-controlled input, creating shell injection (RCE) risks. Additionally, multiple scripts (`scripts/generate_questions.js`, `scripts/screen_resume.js`, `scripts/summarize_interview.js`, `scripts/process_incoming.js`) are vulnerable to Local File Inclusion (LFI) by reading arbitrary files based on user-supplied paths. Most critically, `scripts/screen_resume.js` directly embeds user-controlled arguments (`lang`, `yoe`) into the AI agent's prompt, presenting a clear prompt injection vulnerability that could allow an attacker to hijack the agent's instructions and potentially exfiltrate data or perform unauthorized actions.