docs: add OWASP Agentic Skills Top 10 (AST10) coverage matrix and gap analysis

[grootynix]

Feature Request: Map SkillSpector Detection Patterns to OWASP Agentic Skills Top 10 (AST10)

Summary
SkillSpector's 64 detection patterns across 16 categories provide strong coverage of the attack surface described in the OWASP Agentic Skills Top 10 (AST10). However, there is currently no explicit mapping between SkillSpector rule IDs and AST10 risks. Adding this mapping would:

• Help security teams justify SkillSpector adoption against a recognized framework
• Make scan reports more actionable ("this finding maps to AST04: Insecure Skill Composition")
• Identify any AST10 risks with gaps in current coverage
• Align with how AppSec teams already reason about OWASP frameworks

Proposed Deliverables

1. Coverage matrix — a table mapping each AST10 risk (AST01–AST10) to the SkillSpector rule IDs that address it, with a coverage assessment (full / partial / gap)
2. Rule metadata — add an optional owasp_ast10 field to each pattern definition so it surfaces in JSON/SARIF output
3. Gap analysis — document any AST10 risks not currently covered and propose new patterns or LLM prompts to close them
4. Docs page — add a docs/owasp-mapping.md that can be referenced in reports and READMEs

Reference

• OWASP Agentic Skills Top 10: https://owasp.org/www-project-agentic-skills-top-10/
• GitHub: https://github.com/OWASP/www-project-agentic-skills-top-10
• Related: OWASP Top 10 for Agentic Applications (ASI01–ASI10) — a complementary but distinct framework covering broader agent behavior

Initial Draft Mapping (for discussion)

AST10 Risk	SkillSpector Rules (candidate)	Coverage
AST01 – Malicious Skill Distribution	SC3, SC4, SC6, YR1, YR2	Partial
AST02 – Unsafe Execution Context	AST1–AST8, TT5	Strong
AST03 – Excessive Permissions	PE1–PE3, LP1–LP4, EA1	Strong
AST04 – Insecure Skill Composition	SC1, SC2, SC5, TM2	Partial
AST05 – Prompt Injection via Skills	P1–P5, MP1–MP3, TP1–TP4	Strong
AST06 – Data Exfiltration	E1–E4, TT3, TT4	Strong
AST07 – Insufficient Sandboxing	AST4, AST5, EA2, EA4	Partial
AST08 – Skill Identity Spoofing	TR2, TP2	Weak
AST09 – Lack of Audit & Governance	OH1–OH3, P6–P8	Partial
AST10 – Missing Skill Metadata Standard	LP3, LP4	Weak

The table above is a starting point. I may have the AST IDs slightly off and some rules could map to multiple risks. Happy to collaborate on a more precise version.

Willingness to Contribute
I'm happy to draft the docs/owasp-mapping.md if the maintainers can confirm the intended AST10 risk numbering and whether rule metadata extension is in scope.