Scan for secrets and audit environment permissions for autonomous AI agents.
npx skills add https://github.com/0xAxiom/axiom-public --skill agent-security
Please help me install this Agent Skill. Skill: agent-security Source: https://github.com/0xAxiom/axiom-public/tree/main/agent-tools/skills/agent-security Command: npx skills add https://github.com/0xAxiom/axiom-public --skill agent-security
This Skill provides security guardrails, self-audit tooling, and secret scanning to protect autonomous AI agents from credential leakage, unauthorized actions, and data exfiltration risks.
Copy this skill into the agent's skills directory and run the audit and secret scan:
High-intent search queries and answers about installing and using this skill.
Credential leakage in autonomous agents is prevented through guardrails that enforce non-disclosure policies, secret scanning that detects exposed credentials in workspaces and git history, and environment-based safety checks that block unauthorized data exfiltration before it occurs.
Security guardrails for AI agents enforce non-negotiable policies: prevent credential leakage, block untrusted code execution, reject prompts from untrusted sources, and require explicit human approval for sensitive actions, protecting against unauthorized behavior and data risks.
Self-audit tools verify environment permissions, credential handling practices, and transaction safety by running built-in checks that report on guardrail compliance, secret exposure, and permission misconfigurations across the agent's workspace.
Secret scanning detects credentials and sensitive data in current workspaces and git history, then provides remediation guidance to remove or rotate exposed secrets and prevent future leakage.
These guardrails and audit tools apply to Moltbot and OpenClaw agents, and generalize to any autonomous AI agent requiring hardening, credential protection, and security auditing.
kivo360Configure Claude Code settings.json with permission controls and environment variables.
JaviMontanoDetect and mask exposed credentials across code, configs, docs, and pipelines.
GoldenWing-360Detect leaked credentials in git history and local files.
sjinksReview Infrastructure as Code security controls for privilege, isolation, and deployment weaknesses.
