OpenClaw Shows Why Agents Need a Bill of Materials
Renewed discussion of OpenClaw's local-agent takeover risk shows why teams need an Agent Bill of Materials for skills, plugins, MCP servers, and connectors.
SkillSafe scans, verifies, and delivers AI agent skills to Claude Code, Cursor, Windsurf, Codex, and every tool your team uses. Scanning is free for everyone — no signup required.
Building with a team? Set up your org Why SkillSafe?Prefer a desktop app? Get the SkillSafe desktop app for macOS, Windows, and Linux.
In early 2026, security researchers confirmed 1,184+ malicious AI skills across 12 compromised publisher accounts, distributing Atomic Stealer malware, stealing credentials, and exfiltrating data via ClickFix social engineering. SkillSafe was built so this cannot happen here — every skill is scanned before sharing and independently re-verified on install, with cryptographic tamper detection between publish and download.
Scanning a skill at install time tells you it was clean when you downloaded it. It doesn't tell you it's the same file the publisher uploaded — or that critical findings will actually stop the install.
open registries that scan at install time
Skills are scanned automatically when viewed. AI review classifies findings and flags false positives. Tampered skills are detected via cryptographic tree hashes.
Every archive is SHA-256 hashed into an immutable tree hash. If a single byte changes, verification fails with a critical verdict.
Strict semver enforcement and version yanking. Install exactly the version you need, every time.
Share skills via revocable links with expiration controls. Private by default — share when you're ready, revoke anytime.
Works with Claude Code, Cursor, Windsurf, Codex, Gemini, OpenCode, OpenClaw, Cline, Roo, Goose, GitHub Copilot, Kiro, Trae, AMP, Aider, VS Code, Antigravity, Droid, and Kilo Code. One registry across all your AI agents — no vendor lock-in, no separate extensions needed.
Save up to 50 private skills per day for free. Share with cryptographic verification at no cost. Upgrade only when you need more storage or team features.
Submit a GitHub URL or visit a skill page. SkillSafe runs 12 security checks automatically — no installation required. Results are cached and updated when files change.
AI classifies each finding — separating real threats from false positives. Assigns a security score and produces a recommendation with reasoning for every finding.
Cryptographic tree hashes (SHA-256) ensure files haven't been tampered with. Every version is immutable — any change produces a different hash.
Renewed discussion of OpenClaw's local-agent takeover risk shows why teams need an Agent Bill of Materials for skills, plugins, MCP servers, and connectors.
Cisco's Cloud Control launch puts AI agents, MCP connectors, and third-party tool marketplaces inside critical infrastructure operations.
Vercel opened the skills.sh API for programmatic access to 600,000+ agent skills. That is useful, but it makes skill verification a platform concern.
Organization workspaces are available today: a private shared registry, team roles and invites, domain verification, and security scanning on everything your developers install. Enterprise plans add shared storage, higher limits, and priority support.
See pricing for plan details.
Download the SkillSafe desktop app or run npx skills add to install your first verified skill in seconds — every skill is security-scanned and cryptographically verified.