docs: detection recipes guide

[alexgarden-mnemom]

Summary

• Adds guides/detection-recipes.mdx — public-safe guide covering the three-tier detection architecture, recipe lifecycle, YAML schema reference, and AIP integration
• Updates docs.json to include the new page in the Guides navigation group

Content

The guide covers:

• What detection recipes are (the antivirus analogy)
• Three-tier architecture: Tier 1 heuristics → Tier 2 conscience augmentation → Tier 3 cross-validation
• Recipe lifecycle: draft → active → inactive → archived
• YAML schema reference with a sanitized example recipe
• Integration with AIP (recipes augment, not replace, conscience values)

Does not include: specific agent names/techniques, real thresholds, exact production prompt text, detection rate metrics, or internal model details.

🤖 Generated with Claude Code