Training Manager

• Verdict: Pass
• Confidence: High
• Analyzed: Feb 16, 2026, 5:27 AM
• Findings: 0
• Version: 1.0.2

Overview

The skill's requested files, binaries, and runtime instructions align with its stated purpose of managing an OpenClaw training workspace; the package includes sensible input validation and prompt-injection defenses and does not ask for unrelated credentials or external downloads.

This skill is internally coherent and implements reasonable safeguards (input validation, prompt-injection pattern blocking) for generating and logging training content. Before installing: 1) Be careful not to store secrets (API keys, passwords) inside SOUL.md, MEMORY.md, daily logs, or training entries — logs are written in plaintext. 2) Review any generated SKILL.md or scaffolded files before using them. 3) Backups are created under ~/.openclaw/backups as unencrypted tarballs; if that is sensitive for you, change the backup location or encrypt archives. 4) The skill can autonomously run when invoked by the agent (normal platform behavior); if you want tighter control, only invoke it manually and review outputs. Overall the package appears to be what it claims; proceed if you accept the local filesystem write behavior and the plaintext backups.