Web Researcher
AuditedWeb scraping and summarization Agent with guardrails against privilege escalation.
Security boundaries hand-audited
We hand-audit high-value OpenClaw skills and ship simplified docs
plus an interactive config generator—get a reliable local Agent toolbox in minutes.
Start from these templates—copy config and run locally.
Web scraping and summarization Agent with guardrails against privilege escalation.
Security boundaries hand-audited
Static analysis and security checks for private codebases.
Read-only · no write access
Minimal-privilege execution environment for local scripts.
Path whitelist and param checks
Fill the form on the left; config.jsonupdates live on the right. Copy or download into your OpenClaw project.
Never sent to any server; used only to generate local config.
Comma-separated; helps you identify this config later.
Name this Agent and describe its style and role.
More conservative = safer, less flexible.
Avoid stuck calls; 60–180 recommended.
{
"runtime": {
"provider": "openai",
"apiKey": "<your API key>",
"model": "gpt-4.1-mini",
"timeoutSeconds": 120
},
"agent": {
"name": "ClawLab Operator",
"persona": "Security and observability first; orchestrates audited local OpenClaw skills and avoids privilege escalation or dangerous commands.",
"keywords": [
"openclaw",
"local agent",
"security",
"audited"
],
"safetyLevel": "balanced"
},
"skills": {
"allowed": [
"web-researcher",
"repo-auditor",
"local-runner"
],
"auditing": {
"enabled": true,
"logFile": "./logs/clawlab-audit.log"
}
}
}