Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Hex-level file scanner that detects hidden payloads, control characters, and encoding-based attacks in skill packages. Analyzes binary content that text-based reviewers miss, assigning risk levels and returning structured flags per file.
Text-based code review cannot see null bytes or Unicode directional overrides; hex-level scanning catches injection attempts that standard linters miss entirely.
When to use it
- Auditing downloaded skill packages before running them
- Detecting Unicode directional override attacks in source files
- Scanning an agent tool directory for hidden binary injections
- Verifying file integrity against stored checksums
- Identifying Base64-encoded or obfuscated content in JavaScript files
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: scan this downloaded skill directory for suspicious content before I use it
- 1Call scanDirectory() on the target skill folder
- 2Collect per-file results including riskLevel and flags arrays
- 3Identify any files rated HIGH risk
- 4For each HIGH risk file, retrieve the hexDump and specific flag reasons
- 5Report findings to the user with recommended next steps for manual review
Per-file risk report listing flags such as NULL_BYTES or UNICODE_OVERRIDE, with hex dump excerpts for any HIGH risk files