Skip to content

feat: Phase 5 — Security & Extensibility (s13/s14/s17)#261

Open
Bill-Billion wants to merge 1 commit intoshareAI-lab:mainfrom
Bill-Billion:feat/phase5-security-s13-s14-s17
Open

feat: Phase 5 — Security & Extensibility (s13/s14/s17)#261
Bill-Billion wants to merge 1 commit intoshareAI-lab:mainfrom
Bill-Billion:feat/phase5-security-s13-s14-s17

Conversation

@Bill-Billion
Copy link
Copy Markdown

@Bill-Billion Bill-Billion commented Apr 29, 2026

Summary

  • 3 new chapters: s13 (Permission Guard), s14 (Security Classifier), s17 (Secure Extension Harness)
  • Backend: standalone Python agents (~230/280/450 lines) with tri-lingual docs (en/zh/ja)
  • Frontend: animated SVG visualizations (Learn tab), execution flow diagrams (Deep Dive tab), scenario data (Simulate tab)

Key Changes

New Content

  • agents/s13_permission_guard.py — 5-mode permission system (allow/deny/ask/edit/auto_edit)
  • agents/s14_security_classifier.py — Two-layer classification (regex quick-scan + LLM)
  • agents/s17_secure_extension_harness.py — 5-layer execution pipeline (PreHook → Classify → Permission → Execute → PostHook)
  • Tri-lingual documentation in docs/{en,zh,ja}/

Frontend

  • Animated visualization components: s13-permission-guard.tsx, s14-security-classifier.tsx, s17-secure-extension-harness.tsx
  • Execution flow diagrams for s13/s14/s17 Deep Dive tab
  • Scenario & annotation data for Simulate tab

Improvements

  • Bezier curves in execution flow diagrams — replaced L-shape polylines with cubic bezier curves, eliminating edge crossings and overlaps (benefits all chapters, especially s12)
  • Loop-back routing — backward edges now route along the left side at x=20
  • Same-row connections — horizontal side-to-side routing for nodes on the same y
  • Sidebar — added cyan dot color for "Security & Extensibility" layer
  • i18n — unified zh.json sidebar titles to English for consistency

Test plan

  • python agents/s13_permission_guard.py runs independently
  • python agents/s14_security_classifier.py runs independently
  • python agents/s17_secure_extension_harness.py runs independently
  • npx next build in web/ compiles with zero errors
  • s13/s14/s17 Learn tab shows animated visualizations
  • s13/s14/s17 Deep Dive tab shows execution flow diagrams
  • s13/s14/s17 Simulate tab loads scenario data
  • Execution flow diagrams for s01-s12 have no edge crossings (bezier improvement)
  • Sidebar dot colors display correctly for all layers including security
  • Chinese sidebar shows all titles in English consistently
Add three new chapters to the learn-claude-code curriculum:

- s13: Permission Guard — 5-mode permission system (allow/deny/ask/edit/auto_edit)
- s14: Security Classifier — two-layer pipeline (regex quick-scan + LLM classify)
- s17: Secure Extension Harness — 5-layer execution pipeline combining all layers

Backend: standalone Python agents with tri-lingual docs (en/zh/ja)
Frontend: animated SVG visualizations for Learn tab, execution flow diagrams
  for Deep Dive tab, scenario data for Simulate tab

Also includes:
- Rewrite getEdgePath to use bezier curves, fixing crossing/overlapping edges
  in execution flow diagrams (especially s12)
- Add loop-back routing along left side for backward edges
- Add same-row horizontal connection support
- Add security layer dot color (cyan) to sidebar navigation
- Unify zh.json sidebar titles to English for consistency
@vercel
Copy link
Copy Markdown

vercel Bot commented Apr 29, 2026

@Bill-Billion is attempting to deploy a commit to the crazyboym's projects Team on Vercel.

A member of the Team first needs to authorize it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

1 participant