audit-log-firewall

What This Skill Does

The audit-log-firewall skill serves as a critical security middleware for OpenClaw agents, acting as a gatekeeper for terminal execution. It functions as a dynamic guardrail that prevents unauthorized or dangerous system operations. By enforcing a policy-based interception system, the skill monitors every shell command initiated by the agent. If a command is identified as high-risk—such as destructive filesystem operations, privilege escalation, or unauthorized network requests—the skill triggers an automatic, mandatory 'Pause and Ask' state. This forces the agent to await human authorization before proceeding, effectively mitigating risks associated with autonomous execution. Beyond simple prevention, it maintains a robust forensic trail, hashing every command and logging it into a structured JSON database that includes timestamps, user contexts, and directory information, facilitating deep security audits.

Installation

To integrate the audit-log-firewall into your environment, use the OpenClaw command-line interface. Run the following command in your terminal:

clawhub install openclaw/skills/skills/balkanblbn/audit-log-firewall

Ensure your agent has the necessary write permissions in the root directory to create the .logs/SECURITY.json file. Once installed, the skill will automatically initialize its monitoring service upon the next agent restart.

Use Cases

• Production Environments: Ensuring autonomous agents do not accidentally run destructive commands like rm -rf on mission-critical servers.
• Security Compliance: Organizations needing to maintain an immutable log of all automated actions for audit and accountability purposes.
• Sandbox Development: Safely testing agent behavior when connected to sensitive data by restricting network egress and file modifications.
• Post-Incident Forensics: Using the stored hashes in the security logs to verify that no tampered commands were injected into the agent's execution history.

Example Prompts

1. "Check the current SECURITY.json file and summarize any commands that triggered an interception in the last hour."
2. "Update the allowlist configuration to permit the specific curl command for our internal API endpoint."
3. "Review the audit log for the last session and generate a report of all high-risk commands the agent attempted."

Tips & Limitations

To maximize the utility of this skill, ensure that the config/allowlist.json is strictly maintained and updated as your agent's scope evolves. Do not include wildcard paths that could be exploited to bypass security checks. A key limitation is that the firewall relies on a local configuration; if the agent is compromised by a sophisticated exploit that circumvents local file permissions, the firewall itself might be bypassed. Always integrate this skill with system-level monitoring (like auditd on Linux) for layered security defense.