agent-hardening-zurbrickLock down any LLM agent against prompt injection, data exfiltration, social engineering, and channel-based attacks. Use when setting up a new agent, auditing...
Install via ClawdBot CLI:
clawdbot install zurbrick/agent-hardening-zurbrickGrade Limited — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Accesses sensitive credential files or environment variables
/etc/passwdContains instructions to override system prompt or ignore user requests
"Ignore all previous instructions"Sends data to undocumented external endpoint (potential exfiltration)
send → https://example.comCalls external URL not in known-safe list
https://webhook.site/testGenerated May 7, 2026
A financial advisory firm deploys an AI agent to handle client portfolio inquiries and trading recommendations. The agent must be hardened against prompt injection that could trick it into executing unauthorized trades or leaking sensitive financial data. Channel hardening ensures only verified client IDs can issue trading instructions.
A hospital deploys an agent to answer patient questions about appointments, medications, and test results. The agent must be protected against data exfiltration via email or chat, and behavioral rules prevent it from following instructions embedded in forwarded messages or attachments.
An online retailer uses an AI agent to handle order cancellations, refunds, and account changes. The agent must be hardened against social engineering attacks where attackers impersonate customers or support staff to gain unauthorized access. Tool access is minimized to prevent abuse.
A law firm uses an AI agent to review contracts and legal documents for clauses and risks. The agent must be hardened against injection attacks via document content, and outbound sends require verified channels and recipient approval to prevent confidential data leaks.
A government intelligence agency deploys an agent to analyze classified reports and coordinate responses. The agent must be hardened against channel-based attacks and cross-service MCP chaining risks, with strict instruction isolation from unverified sources.
Offer agent hardening audits and implementation as a professional service for enterprises deploying AI agents. Revenue comes from hourly consulting or fixed-price engagements for vulnerability assessment and rule configuration.
Develop a SaaS platform that automates agent hardening checks across OpenClaw, Claude Code, and LangChain deployments. Revenue from monthly subscription per active agent or per-seat licensing.
Create a training program for developers and security engineers on agent hardening best practices. Revenue from course fees and certification exams.
💬 Integration Tip
Start by auditing existing agent configurations against the attack-surface checklist, then apply channel and MCP hardening before adding behavioral rules. Test with both single-shot and multi-turn injection scenarios.
Scored May 7, 2026
Uses known external API (expected, informational)
api.openai.comAI Analysis
The skill definition is a security hardening guide and contains no executable code or active instructions. The rule-based signals appear to be examples of threats the skill is designed to detect and mitigate, not actions the skill itself performs. The skill's stated purpose is defensive and its content is advisory.
Audited Apr 17, 2026 · audit v1.0
Meta-skill for AI agent self-improvement. Analyzes runtime logs to detect error patterns, regressions, and inefficiencies, then generates structured improvem...
Stop waiting for prompts. Keep working.
Turn OpenClaw into a learning-loop agent with seeded workspace rules, skill promotion, reflective memory, and proactive maintenance.
Meta-agent skill for orchestrating complex tasks through autonomous sub-agents. Decomposes macro tasks into subtasks, spawns specialized sub-agents with dynamically generated SKILL.md files, coordinates file-based communication, consolidates results, and dissolves agents upon completion. MANDATORY TRIGGERS: orchestrate, multi-agent, decompose task, spawn agents, sub-agents, parallel agents, agent coordination, task breakdown, meta-agent, agent factory, delegate tasks
Complete toolkit for creating autonomous AI agents and managing Discord channels for OpenClaw. Use when setting up multi-agent systems, creating new agents, or managing Discord channel organization.
Billions decentralized identity for agents. Link agents to human identities using Billions ERC-8004 and Attestation Registries. Verify and generate authentic...