Skip to content

Latest commit

 

History

History
163 lines (126 loc) · 8.72 KB

SOURCES-QUICK-REFERENCE.md

File metadata and controls

163 lines (126 loc) · 8.72 KB

Sources Quick Reference

Purpose: Fast lookup for the most commonly referenced Tier A/B sources

For complete source database: See SOURCES.md (1,278 lines, comprehensive)

Top 20 Tier A/B Sources

1. Boris Cherny (Claude Code Creator) - Tier A

Role: Engineering Manager at Anthropic, Claude Code creator Source: Paddo.dev Interview (Jan 2026) Key Patterns: Parallel sessions (5 terminal + 5-10 web), plan mode first, natural language git, verification = 2-3x quality Referenced in: 8+ patterns

2. Anthropic Official Documentation - Tier A

Source: https://code.claude.com/docs/en/best-practices Key Guidance: CLAUDE.md ~60 lines, "Would removing this cause mistakes? If not, cut it.", avoid long slash command lists Referenced in: FOUNDATIONAL-PRINCIPLES, 12+ patterns

3. Long-Running Agent Harness - Tier A

Source: Anthropic Engineering Blog (Nov 2025) Key Insights: External artifacts as memory, one feature at a time, structured task lists Referenced in: long-running-agent.md, documentation-maintenance.md

4. Context Engineering for AI Agents - Tier A

Source: Anthropic Engineering Blog (Sep 2025) Key Insights: Context engineering > prompt engineering, context rot, 54% benchmark gains from scratchpad Referenced in: context-engineering.md, FOUNDATIONAL-PRINCIPLES

5. Advanced Tool Use - Tier A

Source: Anthropic Dev Blog (Nov 24, 2025) Key Features: Tool search (85% token reduction), programmatic calling (37% token reduction), input examples Referenced in: advanced-tool-use.md

6. Agent Evaluation Patterns - Tier A

Source: Demystifying Evals for AI Agents Key Insights: Task-based evals, LLM-as-judge, infrastructure noise quantification Referenced in: agent-evaluation.md

7. Beyond Permission Prompts (Security) - Tier A

Source: Anthropic Engineering Blog Key Insights: OS-level sandboxing, permission prompts vs isolation trade-offs Referenced in: safety-and-sandboxing.md

8. OWASP MCP Top 10 - Tier A

Source: https://owasp.org/www-project-mcp-top-10/ Key Insights: 10 critical MCP security risks, mitigation strategies Referenced in: mcp-patterns.md, safety-and-sandboxing.md

9. Nate B. Jones (Agent Principles) - Tier A

Source: Production AI Engineering Best Practices Key Principles: 6 principles for production reliability, lifecycle-aware context model Referenced in: agent-principles.md, memory-architecture.md

10. IndyDevDan Framework - Tier A

Source: YouTube: How I Actually Use Claude Code (Dec 2025) Key Insights: "Great planning is great prompting", The Big Three framework, principles over tools Referenced in: FOUNDATIONAL-PRINCIPLES, spec-driven-development.md

11. GitHub Spec Kit - Tier A

Source: https://github.com/github/spec-kit Key Methodology: Spec-driven development workflow (Specify → Plan → Tasks → Implement) Referenced in: spec-driven-development.md

12. Recursive Evolution (Google TTD-DR) - Tier A

Source: Test Time Diversity for Reliability (Google DeepMind) Key Algorithm: Multi-candidate generation, judge loop, crossover for self-improvement Referenced in: recursive-evolution.md

13. Get Shit Done (GSD) Orchestration - Tier B

Source: https://github.com/glittercowboy/get-shit-done Key Pattern: Fresh context per subagent, state externalization, orchestrator never does heavy lifting Referenced in: gsd-orchestration.md, framework-selection-guide.md

14. CAII (Cognitive Agent Infrastructure) - Tier B

Source: https://github.com/skribblez2718/caii (Kristoffer Sketch) Key Pattern: 7 fixed cognitive agents vs domain-specific proliferation, Johari Window for ambiguity Referenced in: cognitive-agent-infrastructure.md, johari-window-ambiguity.md

15. MCP vs Skills Economics - Tier B

Source: Tenzir: "We Did MCP Wrong" (Jan 2026) Key Data: Skills 50% cheaper than MCP, production cost comparison Referenced in: mcp-vs-skills-economics.md

16. Agentic Retrieval vs RAG - Tier B

Source: LlamaIndex Blog Key Insight: Dynamic navigation vs pre-computed embeddings, context fragmentation solutions Referenced in: agentic-retrieval.md

17. Recursive Context Management (RLM) - Tier B

Source: arXiv:2512.24601 (MIT CSAIL - Zhang, Kraska, Khattab) Key Concept: Programmatic self-examination vs single forward pass, "paradigm of 2026" Referenced in: recursive-context-management.md

18. Session Learning Patterns - Tier B

Sources: Claude Diary (Lance Martin), Generative Agents paper, Yohei Nakajima (BabyAGI) Key Pattern: Capture corrections from sessions, propose updates to persistent config Referenced in: session-learning.md

19. Progressive Disclosure Architecture - Tier B

Source: Production validation (this repository, 73% token savings measured) Key Pattern: 3-tier architecture (main skill + workflow modules + templates), show less reference more Referenced in: progressive-disclosure.md, 10 example skills

20. Evidence Tier System - Tier B

Source: Production validation (this repository, adapted from research methodology) Dual System: A-D for source quality (primary/secondary/tertiary/opinion) + 1-5 for claim strength Referenced in: evidence-tiers.md, writing/research presets

By Pattern Category

Foundational (2)

  1. Spec-Driven Development: GitHub Spec Kit (Tier A)
  2. Framework Selection: Synthesis of GSD, CAII, Native patterns (Tier B)

Context & Planning (4)

  1. Context Engineering: Anthropic Engineering (Tier A), Nate B. Jones (Tier A)
  2. Memory Architecture: Nate B. Jones lifecycle model (Tier A)
  3. Johari Window: CAII methodology (Tier B)
  4. Project Infrastructure: Anthropic docs + Boris Cherny (Tier A)

Security & Quality (4)

  1. Safety & Sandboxing: Anthropic + OWASP (Tier A)
  2. Agent Evaluation: Anthropic evals blog series (Tier A)
  3. Agent Principles: Nate B. Jones 6 principles (Tier A)
  4. MCP Patterns: OWASP MCP Top 10 + Nate B. Jones (Tier A)

Orchestration (3)

  1. GSD Orchestration: glittercowboy/get-shit-done (Tier B)
  2. CAII: skribblez2718/caii (Tier B)
  3. Recursive Context: MIT CSAIL RLM paper (Tier B)

Extensions (3)

  1. MCP vs Skills: Tenzir production data (Tier B)
  2. Progressive Disclosure: Production validation (Tier B)
  3. Advanced Tool Use: Anthropic Dev Blog (Tier A)

Quick Lookup by Need

I Need... Top Source Tier
Core principles Anthropic Official Docs A
Planning workflow GitHub Spec Kit A
Context management Anthropic Context Engineering Blog A
Security guidance OWASP MCP Top 10 + Anthropic Security A
Orchestration GSD (glittercowboy) or CAII (Sketch) B
Cost optimization Tenzir MCP vs Skills B
Self-improvement Google TTD-DR A
Production reliability Nate B. Jones 6 Principles A
Parallel workflows Boris Cherny Interview A
Evaluation patterns Anthropic Evals Blog A

For detailed citations, methodology, and complete source database: See SOURCES.md

Last Updated: February 2026