Knowledge base oriented to AI+Cybersecurity plan

Date: 2026-08-28 Project: astro-starter-template (second-brain)

Decision

The user’s plan = the Security.txt / AI+Cybersecurity guide (pasted from a previous session). Action: add cybersecurity tracks to the site’s curriculum.

What changed

1. lib/curriculum.mjs — new SECTION_DEF cybersecurity (“AI & Cybersecurity”, tags: cybersec), placed after ai-systems, before the rest.
2. Created 12 node stubs in _stubs/cybersec/ (review-first convention) and adopted all 12 into content/nodes/ as unseen:
   • Concepts: prompt-injection, ai-red-teaming, agentic-ai-security, ai-security-governance, security-operations-center, detection-engineering, threat-intelligence, identity-access-management
   • Tools (from the guide): pyrit, promptfoo, openant, defenseclaw
3. IMPORTANT gotcha: a node with tags [cybersec, ai-systems] lands in the AI Systems section because sectionOf first-match-wins and ai-systems precedes cybersecurity in SECTION_DEFS. Fixed by using cybersec alone on those nodes; cross-links to AI Systems stay as edges.

Result

• 174 nodes, 16 sections, 53 courses. New track: 12 nodes in 5 courses: Prompt Injection (with ai-red-teaming, pyrit, promptfoo, openant), Agentic AI Security (+ governance), SOC (+ detection-engineering, threat-intelligence), DefenseClaw, IAM.
• Build green; statuses unchanged (all unseen, level 0).

Cluster constraint

prompt-injection is the root of 4 requires-edges — adopt/move it together with ai-red-teaming and the tools or the build fails on dangling targets.

Future review sessions

The 12 new nodes will surface in the overdue-review flow once they get reviewed/next_review (they are unseen, so they only need capture/logging).