v1.0.0 — Security scanner with effector-core permission parsing

What's in this release

Static security scanner for AI agent tools. Checks for prompt injection, data exfiltration, and permission drift — without executing any code.

Key capabilities

• Static analysis — scans SKILL.md and code for prompt injection, obfuscated instructions, unscoped network/filesystem access
• Permission drift detection — cross-references declared effector.toml permissions against actual capability behavior
• Uses @effectorhq/core for permission parsing — no more regex-based manifest parsing

Quick start

npx @effectorhq/audit ./my-skill
npx @effectorhq/audit permissions ./my-skill

Full changelog: CHANGELOG.md