Source 1
Model Context Protocol Security Best Practices
Official protocol guidance. Reviewed September 7, 2026. Token audience, confused deputy, SSRF, local execution and scope minimization. Versioned documentation URL.
Verified September 7, 2026
Agent Skill Firewall / sources
Sources and research
Sources support the claims identified on each page. The notes describe the material actually reviewed and significant limits. Interpretations, proposed workflows and planning exercises are our editorial synthesis. No third-party figure, photograph or substantial source text is reproduced.
Source 2
OWASP Agentic Skills Top 10
OWASP project guidance, marked public review v1 at access. Used for skill-specific risk categories, not copied incident totals or claims of universal coverage.
Verified September 7, 2026
Source 3
OWASP Top 10 for Agentic Applications
OWASP project primary publication. Broader agentic application risk and governance context.
Verified September 7, 2026
Source 4
Anthropic on Claude Code sandboxing
First-party implementation discussion. Filesystem and network isolation are implementation examples; vendor efficacy claims are not adopted as guarantees.
Verified September 7, 2026
Source 5
OWASP skill security checklist
Primary project checklist. Additional reading on installation, execution and governance. Our review logic is original, not an OWASP rating or certification.
Verified September 7, 2026
Corrections and maintenance
Recheck material scientific or technical claims when a primary source changes. Product documentation and branch-based repository references may change without a new URL. Send supported corrections through the contact page.
Continue reading