LaunchKit · 2026
Back to Skills

agent-hardening

Test your agent's input sanitization against common injection attacks.

4
649 downloads
by @x1xhlol

Setup & Installation

openclaw skills install @x1xhlol/agent-hardening

Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:

npx clawhub install agent-hardening

Version History

v1.1.2Feb 16, 2026 - Version note by skill developer:

Added back threat definitions reference. All checks remain self-contained with synthetic data.

What This Skill Does

Runs self-contained security checks on OpenClaw agents to test input sanitization. Covers unicode control character stripping, HTML comment injection detection, and bidi override character identification. Uses only hardcoded synthetic samples — no local files or external connections.

Requires nothing beyond Python 3, so checks run immediately with no setup, accounts, or network access.

When to use it

  • Verify an agent strips zero-width unicode characters before processing user input
  • Detect hidden HTML directives injected into agent context windows
  • Check if filename spoofing via bidi override characters is caught
  • Run a quick pre-deployment security audit on a new agent
  • Validate input sanitization logic during agent development

Example Workflow

Here's how your AI assistant might use this skill in practice.

INPUT

User asks: run the agent hardening checks on my agent

AGENT
  1. 1Runs unicode control character sanitization test against synthetic samples
  2. 2Runs HTML comment injection detection against sample strings
  3. 3Runs bidi override detection against sample filenames
  4. 4Reports pass/fail counts for each check category
OUTPUT

Summary showing how many checks passed per category, with flagged samples highlighted