Ai Interview
Security checks across malware telemetry and agentic risk
Overview
The skill is coherent, but its web viewer can expose and delete OpenClaw agent session logs through an unauthenticated network-accessible local server.
Only install or run this if you are comfortable exposing OpenClaw agent session logs to the viewer. If you use it, run the server on a trusted machine, firewall port 8091, avoid opening untrusted webpages while it is running, use dedicated Feishu bot credentials, and back up session logs before using the clear function.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Private conversations or sensitive context from other OpenClaw agents could be displayed in the interview viewer.
The viewer enumerates and parses OpenClaw agent session files from the global agents directory, not just the job-seeker and recruiter sessions.
SESSION_DIR = Path.home() / ".openclaw" / "agents" ... sessions = list(agent_dir.glob("*.jsonl")) ... result['agent-1'] = parse_session_messages(session1)Limit the viewer to explicitly configured interview agents, validate allowed agent IDs, and warn users before reading or displaying persistent session logs.
Other devices on the network, or a webpage running in the user's browser, may be able to read agent session data from the viewer API.
The HTTP server listens on all network interfaces and permits any web origin to read API responses, with no authentication shown.
self.send_header('Access-Control-Allow-Origin', '*') ... server = HTTPServer(('', PORT), Handler)Bind only to 127.0.0.1 by default, require an access token, remove wildcard CORS, and document the network exposure clearly.
A mistaken click, local webpage request, or network request could erase agent conversation history.
A GET API endpoint deletes all .jsonl session files for user-supplied agent names without authentication, CSRF protection, or a server-side confirmation step.
elif path == '/api/clear': ... cleared = clear_conversations(agent1, agent2) ... s.unlink()
Make clearing logs an authenticated POST action, restrict it to known interview agents, add confirmation, and prefer moving files to a recoverable trash location.
The configured bots will be able to read relevant Feishu group messages and send messages as bots.
The skill asks users to configure Feishu app secrets and bot permissions so the two interview agents can send and read group messages.
"appSecret": "你的job-seeker App Secret" ... im:chat:send_as_bot ... im:message:content:readonly ... im:chat:readonly
Use dedicated Feishu apps with minimal scopes, keep secrets out of shared files, rotate secrets if exposed, and only add the bots to intended groups.