saferoactiveA secure, human-approved autonomous agent architecture that combines SMFOI-KERNEL orientation with Write-Ahead Logging (WAL), Proposal-First decision-making,...
Install via ClawdBot CLI:
clawdbot install rigeneproject/saferoactiveGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Contains instructions to override system prompt or ignore user requests
"Your purpose is"Calls external URL not in known-safe list
https://github.com/openclaw/skills/issuesAI Analysis
The skill defines a local security framework with no evidence of data exfiltration or credential harvesting. The primary risk is the 'PROMPT_POISONING' signal, which suggests an attempt to define the agent's purpose, but this appears within the context of a documented security protocol rather than a hidden override. The external URL reference is a low-risk issue tracker link consistent with maintenance.
Audited Apr 17, 2026 · audit v1.0
Generated Oct 8, 2026
An AI agent handles first-level customer support tickets, classifying and routing them to the appropriate department. All actions that create or modify tickets are treated as Level 2 and require human approval, ensuring a human-in-the-loop for critical customer interactions.
In a legal or financial setting, an agent extracts and analyzes information from contracts, generating a proposal for any new file creation. The mandatory WAL and approval gate ensure that no unauthorized documents are generated, satisfying audit and compliance requirements.
An agent answers employee questions by searching internal documentation and creating summary notes. Any new file creation in the knowledge base is Level 2 and requires human approval, preventing accidental data leaks or misinformation.
A clinical support agent helps medical staff retrieve patient information and draft reports. Due to strict privacy regulations (e.g., HIPAA), any action that writes to the system is gated behind manual approval, and filesystem access is confined to designated safe directories.
An AI agent proposes infrastructure changes or new configuration files as part of a CI/CD pipeline. Each proposed change is written to a WAL and requires a human operator's explicit approval before execution, reducing the risk of unintended outages.
Offer SafeProactive as a cloud service where compliance officers can monitor and approve AI actions through a dashboard. The service includes audit logs, role-based approvals, and integration with existing GRC tools.
Sell a self-hosted version of SafeProactive to large enterprises that require strict data sovereignty. The license includes setup, training, and premium support for integrating with internal systems.
Distribute the core framework as open-source under MIT, and monetize through premium extensions such as advanced approval workflows, integration connectors (e.g., Slack, Jira), and enhanced audit reporting.
💬 Integration Tip
Implement SafeProactive by first defining the filesystem boundaries and setting up the WAL directory; then integrate approval gates into your agent's decision loop to ensure human oversight for Level 2+ actions.
Scored Oct 8, 2026
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope,...
Security scanner for AI agent skills. 9 built-in detection signatures. Identifies secrets, unsafe execution patterns, and prompt injection. Sub-50ms results.
Wallet anti-theft guard. One-click scan for high-risk wallet approvals to protect user assets. Use when a user asks for a wallet security check, wallet healt...
Comprehensive security audit for an agent's full skill stack. Chains scanner, differ, trust-verifier, and health-monitor into a single assessment with priori...
GEO Audit — AI Search Visibility Checker for ChatGPT, Perplexity, Claude & Gemini. 29-point GEO readiness checklist: robots.txt AI crawler access, Index...
Senior SecOps engineer skill for application security, vulnerability management, compliance verification, and secure development practices. Runs SAST/DAST sc...