AI from Dynalist - Sorted

Processed from AI_from_Dynalist.md. Notes categorized and lightly enhanced for clarity. Items marked with #action need follow-up. Items marked #eval need evaluation before adopting.

---

Claude Code

CC Workflow Upgrades

• Review and ACT on CC /insights report: file://wsl$/Ubuntu-24.04/home/ta/.claude/usage-data/report.html
• Set monthly task to run CC /insights — report findings, then implement upgrades
• Learn how to use Claude web with CC efforts: clau.de/web

Claude Code SDK

• https://docs.anthropic.com/en/docs/claude-code/sdk
• CC with Cursor, with SDK integration

Setup and Config

• d:\FSS\Websites\monorepo\docs\CLAUDE_CODE_SETUP.md
• d:\FSS\Websites\monorepo\docs\MCP_CONFIGURATION.md
• d:\FSS\Websites\monorepo\docs\PHASE_1_COMPLETE.md
• See agent coordination repo: https://github.com/ilyasibrahim/claude-agents-coordination/releases/tag/v1.0.0
  • Balanced: technical correctness, workflow value, performance constraints, maintainability, usability
  • https://medium.com/@ilyas.ibrahim/how-i-made-claude-code-agents-coordinate-100-and-solved-context-amnesia-5938890ea825
• https://alirezarezvani.medium.com/the-only-3-things-you-need-to-start-with-claude-code-everything-else-can-wait-e15d3428d5ef
• Run /init to create a proper CLAUDE.md file — this is programming your AI assistant
• Use npm version of Claude Code (enables Claude Trace): https://github.com/badlogic/lemmy/tree/main/apps/claude-trace
• Max reasoning config in ~/.claude/settings.json:
  • "CLAUDE_CODE_MAX_OUTPUT_TOKENS": "64000", "MAX_THINKING_TOKENS": "31999"

CLAUDE.md and Memory

• CC maintains project MEMORY.md automatically: \\wsl$\Ubuntu-24.04\home\ta\.claude\projects\-home-ta-projects-monorepo\memory\MEMORY.md
• Use # shortcut during sessions to add instant memory updates
• Schedule monthly CLAUDE.md audits — delete stale, add missing
• Add planning instructions to CLAUDE.md (get from https://alirezarezvani.medium.com/your-claude-md-is-probably-wrong-7-mistakes-boris-cherny-never-makes-6d3e5e41f4b7)
• Add verification requirements to CLAUDE.md
• CLAUDE.md best practices:
  • Keep under 100 lines (context efficiency)
  • Add subdirectory CLAUDE.md for module-specific rules
  • Use absolute paths
  • CLAUDE.md as a roadmap to documentation, not the documentation itself — use @path/to/file imports
• Four memory scopes: Enterprise, Project, Sub-folder, Personal
  • Personal: ~/.claude/CLAUDE.md — preferences across all projects
  • Enterprise always highest priority
• Hot reload only works for skills; CLAUDE.md changes still require session restart
• Memory solutions: https://medium.com/@me_82386/i-lost-47-claude-code-conversations-before-building-this-47995856a283
• Good memory strategies: https://medium.com/@luongnv89/claude-code-memory-teaching-claude-your-projects-dna-45c4beca6121

Advanced Strategies

• Extended Thinking levels: “think” (moderate), “think hard” (challenging), “think harder” (architectural), “ultrathink” (debugging hell)
• Checkpoints and /rewind for fearless experimentation
• Subagents: parallelize dev by spawning specialized agents (backend, frontend, tests)
• Context Hygiene: /clear to maintain clarity
• Planning Mode First: separate planning from coding (Shift+Tab twice)
• Custom Slash Commands: automate repeated workflows
• VS Code Extension: blend CLI and IDE workflows
• Hooks: post-edit and pre-commit scripts for code safety
• Human-Guided Git: manual control over Git history
• Use Haiku 4.5 for simpler tasks (much cheaper)
• Haiku is the sprinter, Sonnet is the steady builder, Opus is the careful reviewer
• Model switching: Alt+P (Windows) mid-session
• Name sessions for easy resume
• Start new chats after small steps to clear context window
• Use PLAN MODE first for everything non-trivial, then auto-accept edits mode
• Content in projects is cached and doesn’t count against limits when reused
• /statusline: useful for model and token usage info
• Tool Search: lazily loads MCP tools, cutting context waste
• Async hooks: run in background, ~3x faster workflows
• Learn about Project mode: claude --project

Skills and Slash Commands Merged

• Skills gained ability to be invoked with /skill-name
• Simple commands: single markdown in .claude/commands/
• Complex skills: directory structure in .claude/skills/skill-name/ with multiple files, scripts, templates
• Hook scoping hierarchy: global -> project -> skill -> sub-agent
• Skills auto-reload when SKILL.md changes

Plugins

• Combine /commands with MCP servers; marketplace available
• Plugin system: custom commands, agents, hooks, MCP servers
• CCPlugins: 24 predefined commands extending CC CLI

Frameworks and Enhancements

• Superpowers plugin: https://github.com/obra/superpowers — highest-ROI action #eval
  • Adds /brainstorm and /write-plan commands
  • https://medium.com/vibe-coding/every-ai-tool-has-plan-mode-none-of-them-do-it-right-6bd540155690
• SuperClaude #eval
• “Everything Claude Code” — production code dev system, use as plugin/MODEL: https://medium.com/@joe.njenga/everything-claude-code-the-repo-that-won-anthropic-hackathon-33b040ba62f3
• Claude Tasks: https://medium.com/@joe.njenga/claude-code-tasks-are-here-new-update-turns-claude-code-todos-to-tasks-a0be00e70847
• Auto Claude: autonomous multi-agent framework: https://github.com/AndyMik90/Auto-Claude
  • https://medium.com/@joe.njenga/i-tested-this-autonomous-framework-that-turns-claude-code-into-a-virtual-dev-team-a030ab702630
• GSD (Get Shit Done): spec-driven framework for CC: https://github.com/glittercowboy/get-shit-done
  • https://medium.com/@joe.njenga/i-tested-gsd-claude-code-meta-prompting-that-ships-faster-no-agile-bs-ca62aff18c04
• Claude-Flow v2.0: enterprise-grade AI orchestration: https://github.com/ruvnet/claude-flow
• Multi-agent with tmux: https://javascript.plainenglish.io/i-built-an-ai-team-that-codes-itself-and-you-can-too-974a48b4ef49

Claude Artifacts

• https://kotrotsos.medium.com/claude-artifacts-the-features-that-replace-500-month-in-software-a44388375280
• Explore for micro apps, interactive AI-driven courses

Learning Resources (CC-specific)

• CC Secret Weapon (Max Petrusenko): https://medium.com/dare-to-be-better/claude-code-the-ai-developers-secret-weapon-0faac1248080
• Claude Code Documentation
• Anthropic Best Practices
• Free course: anthropic.com/learn (“Claude Code in Action”)
• Deep Learning AI course: https://learn.deeplearning.ai/courses/claude-code-a-highly-agentic-coding-assistant/lesson/66b35/introduction
• Who to follow: https://jpcaparas.medium.com/who-to-follow-if-youre-serious-about-claude-code-0d49abe2d521
• CC Cheatsheet: https://awesomeclaude.ai/code-cheatsheet
• The Only CLAUDE.md Guide: https://medium.com/@aleksandardobrohotov/the-only-claude-md-482b771431b2
  • Domain Rule: files that change together live together
  • Name Rule: file names must be self-explanatory
  • Absolute paths or death
• Boris Cherny tips: https://medium.com/@joe.njenga/boris-cherny-claude-code-creator-shares-these-22-tips-youre-probably-using-it-wrong-1b570aedefbe
• 10-Level CC Cheat Sheet: https://medium.com/@joe.njenga/my-10-level-claude-code-cheat-sheet-will-make-you-a-pro-in-minutes-which-level-are-you-54dd7e155140
• 47 CC WSL Tricks: https://medium.com/@joe.njenga/47-claude-code-wsl-tricks-every-windows-user-should-know-5d42aaee2d93
• 25 CC Setup Steps: https://medium.com/@joe.njenga/25-claude-code-set-up-steps-for-pro-devs-did-you-miss-a-step-ad57ee924554
• CC 2025 Summary: https://medium.com/@joe.njenga/claude-code-2025-summary-from-launch-to-beast-timeline-features-full-breakdown-45e5f3d8d5ff
• CC 2.1 Guide: https://medium.com/codex/claude-code-21-guide-16e084fd5be0
• CC 2.1 time-saving features: https://alirezarezvani.medium.com/these-4-claude-code-2-1-features-save-me-at-least-50-minutes-daily-95-developers-skip-them-46d2d10f03cc
• CC with OpenRouter low cost: https://medium.com/@joe.njenga/how-im-using-claude-code-like-cline-with-openrouter-to-go-beast-mode-at-low-cost-8c78e0bdcb67
• CC Max Reasoning: https://medium.com/coding-nexus/how-to-run-claude-code-with-opus-4-5-at-maximum-power-525cdc1c470e
• CC with GLM-4.7: https://medium.com/@joe.njenga/i-tried-claude-code-with-glm-4-7-heres-what-you-are-missing-453edb4424d3
• Open-sourced code cleaner: https://medium.com/coding-nexus/anthropic-just-open-sourced-a-code-cleaner-that-actually-gets-it-e2a6eeea70da
• My CC Workflow for 2026: https://youtu.be/sy65ARFI9Bg?si=agf4FNMY4B0GpJXb
• CC in Production: https://medium.com/coding-nexus/how-to-use-claude-code-to-the-fullest-for-developers-in-production-611b1e3ea617
• 30 CC Subagents with templates: https://alirezarezvani.medium.com/30-claude-code-subagents-you-need-in-2026-with-templates-you-can-use-today-part-1-2-127a11f473b1
• Skills Factory: https://alirezarezvani.medium.com/the-claude-skills-factory-how-you-can-generate-production-ready-ai-tools-in-15-minutes-eb0b86087f31
• CC vs Google CLI comparison: https://www.perplexity.ai/search/compare-cloud-code-and-google-TOL3shifS.aJjpFHjK1A8Q
• Clawdbot: for devs and more #eval
• Vibe Engineering 24/7: https://medium.com/@dzianisv/vibe-engineering-how-to-keep-ai-coding-tools-like-claude-code-running-24-7-on-your-behalf-787aa89cb81b
• SkillsMP 87K+ directory article: https://medium.com/ai-software-engineer/skillsmp-this-87-427-claude-code-skills-directory-just-exploded-but-with-one-annoying-problem-ec4af66b78cb
• CC Cheat Sheet (ToniMaxx): https://medium.com/@tonimaxx/the-ultimate-claude-code-cheat-sheet-your-complete-command-reference-f9796013ea50

CC vs Google CLI Strategy

• Start with Gemini’s free tier for prototyping, switch to Claude for production reliability
• Gemini for 80% of daily tasks (scaffolding, Google integration, deployment)
• Claude for 20% that matters most (business logic, security, complex features)

---

Cursor

Cursor 2.0

• https://cursor.com/changelog/2-0
• Plan mode: structured execution plans
• Composer: new interface for multi-agent parallel work
• Browser control for enhanced web dev
• Hooks (beta) for additional functionality
• Voice mode
• CLI: https://cursor.com/docs/cli/overview

Cursor Config

• Rules and Memories: system-level instructions for Agent and Inline Edit
• Custom commands: reusable workflows with / prefix
  • Project: .cursor/commands/
  • Global: ~/.cursor/commands/
• Visual Designer via Browser (icon at top)

Cost Optimization

• Subscription: US $20/mn (as of 2025OC25)
• Could use cheaper models in Cursor very cost effectively:
  • Gemini 2.5 Pro (free?)
  • Claude Haiku 4.5
  • Better models when needed

---

Skills

Skills Philosophy

• Skills are a significant advance in AI effectiveness — as significant as spreadsheets for productivity
• Accessible: process + knowledge + resources + code functionality, all in English
• No coding needed, no APIs needed — just clear plain text instructions
• Validated by how quickly OpenAI created their own skills capability after Anthropic
• A skill is version-controlled, auditable, testable, and attributable
• Simple structure creates a compounding flywheel of knowledge and capabilities
• A skill can include templates, configuration, scripts, and reference documentation in one package
• Published as open standard at agentskills.io

Skills Structure

• Folders containing SKILL.md file, scripts, and optional resources
• Portable, composable, efficient — Claude dynamically loads only relevant skill
• Use gerund form (verb + -ing) for naming: processing-pdfs
• Description should state what the Skill does AND when to use it
• !command syntax injects shell command output into skill content
• context: fork directive runs skill in isolated subagent
• PreToolUse, PostToolUse, and Stop hooks can be defined in skill frontmatter

Skills Best Practices

• Official guide: https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices
• Complete guide PDF: https://resources.anthropic.com/hubfs/The-Complete-Guide-to-Building-Skill-for-Claude.pdf?hsLang=en
• Skill-creator SKILL.md reference: https://github.com/anthropics/skills/blob/main/skills/skill-creator/SKILL.md
• Progressive Disclosure architecture: only load heavy context when necessary, keeping context window cheap and fast
• Supercharging front-end dev with skills: https://dev.to/rio14/supercharging-front-end-development-with-claude-skills-22bj

Skills Meta-Tools

• Skill-Creator (Anthropic Official): step-by-step skill creation, init_skill.py, package_skill.py
  • URL: https://github.com/anthropics/skills/tree/main/skills/skill-creator
  • Install: /plugin marketplace add anthropics/skills then select skill-creator
  • Use /skill-creator to build skills
• Skill-Updater: systematic 5-step update process
  • URL: https://mcpmarket.com/es/tools/skills/skill-updater
• Awesome-Claude-Skills (Community): curated open-source skills
  • https://github.com/VoltAgent/awesome-claude-skills
  • https://github.com/ComposioHQ/awesome-claude-skills
• Skills Factory (CCGG community): generates complete skills from brain dumps
  • https://ccggskool.com
• Skilz Universal Installer: works with 14+ coding agents
  • https://skillzwave.ai/docs/

Skills Repos and Directories

• https://github.com/anthropics/skills/tree/main
• https://github.com/ComposioHQ/awesome-claude-skills (includes skill-creator)
• https://jimmysong.io/ai/awesome-claude-skills/
• https://awesomeclaude.ai/
• https://skillsmp.com/docs/api (87,000+ skills directory)
  • https://medium.com/@jh.baek.sd/i-found-120-000-ai-experts-who-work-for-free-heres-how-to-use-them-1c22aff6c5d8

Skills Articles

• “The AI Agent Race is Over. The Winner is a Folder.”: https://kotrotsos.medium.com/the-ai-agent-race-is-over-the-winner-is-a-folder-8cdc7ad7bbb2
• “Don’t Build Agents, Build Skills Instead”: https://youtu.be/CEvIs9y1uog?si=9xphfFOsKmp7-Ca3
• Self-improving Skills in CC: https://youtu.be/-4nUCaMNBR8?si=S76Py4NLXJY3CdUc
• “Quit Prompting, Start Building”: https://kotrotsos.medium.com/quit-prompting-start-building-e678b12c5656
• Mastering Agentic Skills (Spillwave): https://pub.spillwave.com/mastering-agentic-skills-the-complete-guide-to-building-effective-agent-skills-d3fe57a058f1
• Claude Skills changed how we use AI: https://medium.com/coding-nexus/claude-skills-just-changed-how-we-use-ai-29b6bab97be5
• How Agent Skills became AI’s most important standard: https://ai.gopubby.com/how-agent-skills-became-ais-most-important-standard-in-90-days-a66b6369b1b7
• Master CC Skills tool: https://medium.com/nginity/master-claude-code-skills-tool-to-transform-repetitive-ai-prompts-into-permanent-executable-5dc9a4711f28
• MCP vs Skills: MCP defines how Claude calls external systems; Skills define what Claude should do
• Skills and Slash Commands merged: https://medium.com/@joe.njenga/claude-code-merges-slash-commands-into-skills-dont-miss-your-update-8296f3989697
• Build Agent Skills Faster (CC 2.1): d:\FSS\KB\Business\Clippings\Build Agent Skills Faster with Claude Code 2.1 Release.md

Specific Skills to Install/Build

• Playwright Skill: E2E testing: https://github.com/lackeyjb/playwright-skill
• frontend-design (Anthropic): modern UI design principles, prevents “AI Slop”
• Superpowers (Obra): /brainstorm and /write-plan
• ffuf-web-fuzzing: AI security auditor
• SvelteKit component generators, Tailwind stylers, debt calculator APIs
• Strategic decisions skill (see Rick Moretti YT)

Skills for SMART DEBT Mission

• Web dev skills (SvelteKit refactors, debugging, API mocks)
• Branding/content generation
• Financial education app prototypes
• Smart debt coach automated service

---

Obsidian (KB)

Obsidian Skills

• Official Obsidian Skills: https://kurtis-redux.medium.com/obsidians-official-skills-are-here-it-s-time-to-let-ai-plug-into-your-local-vault-6c149aae84f6

---

Web Dev Tools

Browser Automation

• Playwright: primary browser automation
• Agent-Browser (Vercel Labs): https://skillzwave.ai/agent-skill/vercel-labs-agent-browser-agent-browser/
  • https://medium.com/@richardhightower/agent-browser-ai-first-browser-automation-that-saves-93-of-your-context-window-7a2c52562f8c
• Browser-Use + CrewAI: max flexibility for local agent system
  • https://github.com/browser-use/browser-use
  • https://github.com/crewAIInc/crewAI
• Nanobrowser: zero-setup FOSS browser automation
  • https://github.com/nanobrowser/nanobrowser
• Claude can use Chrome extension to control browser

MCPs (Model Context Protocol)

• Chrome DevTools MCP
• GitHub MCP Server: https://github.com/github/github-mcp-server
• GitMCP: https://github.com/idosal/git-mcp
• Filesystem MCP: https://github.com/modelcontextprotocol/servers/tree/main/src/filesystem
• Context7 MCP: https://github.com/upstash/context7 — up-to-date docs for LLM use
• Sequential Thinking MCP: structured problem-solving — https://github.com/modelcontextprotocol/servers/tree/main/src/sequentialthinking
• Memory MCP: knowledge graph persistence — https://github.com/modelcontextprotocol/servers/blob/main/src/memory
• 10 open-source MCPs: https://medium.com/@devlink/10-open-source-mcps-that-make-your-ai-agents-smarter-than-your-team-lead-02283f0bd941
• MCP Toolbox for Databases: https://googleapis.github.io/genai-toolbox/getting-started/introduction/
• Playwright MCP: browser automation
• Browser MCP: browser automation with coding agents
• Firecrawl MCP: scrape, crawl, search, extract
• Magic MCP: modern UI components from IDE
• Lingo MCP: AI localization toolkit
• Sentry MCP: bug analysis and fix in one flow
• Markitdown: https://github.com/Klavis-AI/klavis/tree/main/mcp_servers/markitdown
• Windows CLI MCP: https://github.com/SimonB97/win-cli-mcp-server
• Mem0 MCP: memory layer for AI agents #eval
• Memory MCP: knowledge graph persistence
• Google Calendar MCP: https://github.com/nspady/google-calendar-mcp

MCP Marketplaces

• https://github.com/modelcontextprotocol/registry
• https://smithery.ai/
• https://mcpservers.org
• Cline has extensive marketplace

Spec Dev Alternatives

• GSD (Get Shit Done): https://github.com/glittercowboy/get-shit-done — simpler #eval
• GitHub Spec Kit: https://youtu.be/a9eR1xsfvHg?si=nWnmHmbIShv4yUCc
• Traycer.ai: production quality agentic planning

Dev Tools

• For Svelte: SvelteGPT #explore, SvelteKit Sensei #explore
• Superflex: VS Code extension for frontend dev
• Snyk: security checking
• Jam: AI debugging with screen recording
• Mintify: AI generated code documentation
• Polypane: browser for responsive design, accessibility, performance
• Gitdocs AI: https://www.gitdocs.space/ (README generation)
• DevToys: offline Swiss-army tool (JWT, regex, color, diffs)
• OpenHands: AI agent for commands, edits, tests

Coding Practices

• https://medium.com/h7w/9-ai-powered-coding-practices-that-turn-average-developers-into-senior-level-performers-0cd428d7e61d

Agentic IDEs

• https://www.builder.io/blog/agentic-ide (best agentic IDEs heading into 2026)

Dev UI References

• Webflow Templates for UI: https://webflow.com/templates

---

AI Dev (Workflow, Upgrades)

AI Dev Workflow Core

• See: _Working On/AI Dev Workflow.md
• https://medium.com/vibe-coding/someone-finally-built-the-ralph-loop-i-wanted-8f3050b7b181 (Ralphy upgraded structure)
• Before automating: use AI to generate markdown specs, run tasks manually once, refine prompts, then wrap in a loop
• Each loop: one task -> finish -> commit -> update plans -> exit for next loop
• Use Ralph only where completion criteria are clear and verifiable, set strict spending limits
• Get another AI model to review results from the first; for critical issues, use best models + second opinions
• At end of each task, ask AI to review for lessons learned
• 59% of developers run 3+ AI tools in parallel
• Use Haiku 4.5 for implementation (much cheaper)
• “Asynchronous development”: describe feature before bed, AI builds overnight, review in morning

Planning First

• Start with clear PRD (Product Requirements Doc) including sequence diagrams
• Plan first, code later — prevents scope creep
• Use structured docs before opening Cursor
• Build UI first, then development
• Combine tools: Cursor + CodeGuide + Lovable = ultimate speed
• Test thoroughly, commit after each confirmed step
• Use Code Rabbit for testing?

Multi-Tool Strategy

• GLM-4.7: ~Claude-level quality at fraction of cost ($3-6/month)
• Start with Gemini free tier for prototyping, Claude for production
• Multi-prompting: send same prompt to 3 models, synthesize best insights
• 80/20 rule: Gemini for daily tasks, Claude for what matters most
• OpenRouter: access any LLM from single connection

AI Dev Workflow (“I’ve Built 50+ Apps” Model)

• https://youtu.be/99FI5uZJ8tU?si=Wk7amf1SX9ySFvNZ
• Excellent AI dev workflow: MODEL this #action

Prep Phase

• Ask AI for suggestions on best way to proceed
• Ask AI to improve initial project outline
• Ask if AI completely understands the project, what could be clarified
• ChangeLog.md: always have AI write changes there

Claude Tasks

• https://medium.com/@richardhightower/claude-code-todos-to-tasks-5a1b0e351a1c
• https://medium.com/@joe.njenga/claude-code-tasks-are-here-new-update-turns-claude-code-todos-to-tasks-a0be00e70847

Gemini Code Assist (distinct from CLI)

• https://codeassist.google/
• Google Cloud Shell Editor: 50 hours/week free with Gemini Code Assist pre-installed
• Use Google AI Studio free reasoning with API key inside Cursor: https://youtu.be/6h9y1rLem4c?si=sRtdzKMqqlrMrUYT

Gemini CLI Conductor #eval

• Track-to-implementation: spec -> plan -> phases -> tasks -> subtasks
• /conductor:implement follows plan.md with checkpoints
• https://developers.googleblog.com/conductor-introducing-context-driven-development-for-gemini-cli/
• Tested Gemini CLI Conductor: https://medium.com/ai-software-engineer/i-tested-gemini-cli-conductor-new-way-to-code-without-getting-lost-in-ai-chaos-fc4338371679
• GPT analysis chats saved

Vibe-Kanban #eval

• Supercharge CC, Gemini CLI, any AI coder
• Multiple models in parallel
• Unified MCP server configuration

Agentic Development

• Agent-OS (Brian Casels): good project mgmt structure #eval
• DeepCode: agentic coding (Paper2Code, Text2Web, Text2Backend): https://github.com/HKUDS/DeepCode
• Google Gemini agent stack (open source): https://medium.com/coding-nexus/googles-open-source-gemini-agent-stack-build-production-grade-ai-agents-you-can-actually-ship-73d0fb0ee671
• Multi-agent AI army (Abdul Ahad): https://ai.plainenglish.io/the-day-i-built-my-first-ai-agent-army-how-multi-agent-ai-changed-the-way-i-solve-problems-5f76d0a23222
• AI Agile Dev: YouTube “agile coding is here”
• Auto Claude: autonomous multi-agent framework
• Claude-Flow v2.0: enterprise swarm intelligence
• Multi-agent tmux-based approach
• Agentic prompts:
  • “Write step-by-step plan… tell me edge cases”
  • “Now critique this plan. What might go wrong?”
  • “Break into subtasks and spin up sub-agents in parallel”
  • “Double-check everything. What might this have broken?”

Ralph Loop Best Practices

• https://medium.com/ai-software-engineer/7-ralph-loop-mistakes-that-are-burning-your-tokens-and-wasting-your-time-b506c384face

Using AI for Strategic Development

• Laptop as full-time AI developer (Abdul Ahad): https://medium.com/ai-simplified-in-plain-english/how-i-turned-my-laptop-into-a-full-time-ai-developer-and-watched-it-outperform-me-c798012ae4b4
• https://youtu.be/kDcM_xwmP3Q (coded app with AI in 67 mins)
• https://sakkyb.medium.com/how-i-built-a-saas-product-with-100-ai-generated-code-5728e0c97c8d
• https://wire.insiderfinance.io/how-i-built-13-mvps-in-90-days-with-ai-code-5953585a7dfc
• Framework for AI dev: https://medium.com/@aleksandardobrohotov/how-i-let-ai-rebuild-my-website-in-7-days-using-vibe-coding-complete-step-by-step-guide-8407cabeb5d5

---

Resources

AI Learning Platforms

• OpenAI Academy (free): https://academy.openai.com/
• Anthropic Learn: anthropic.com/learn
• Deep Learning AI: https://learn.deeplearning.ai/
• Warp University: https://www.warp.dev/university

Prompting Resources

• Platform.claude.com: https://platform.claude.com/dashboard (prompt generator)
• Promptize: https://promptize.net (optimize prompts)
• CLEAR framework: Context, Length, Examples, Audience, Role
• Role -> Context -> Goal -> Output Rules
• Strip filler (“please”, “thank you”) unless contextually important
• Start with imperative verb: “Analyze…”, “Summarize…”, “Generate…”
• 10 Expert Prompt Templates: https://generativeai.pub/10-expert-prompt-templates-i-use-as-a-developer-and-writer-d6e97275edff
• Ultimate ChatGPT Prompt Guide: https://medium.com/aimonks/ultimate-chatgpt-prompt-guide-for-2024-29ed7add2507
• Perspective Prompting (Michael Simmons): five principles for multi-perspective AI
• “I accidentally made Claude 45% smarter”: https://medium.com/@ichigoSan/i-accidentally-made-claude-45-smarter-heres-how-23ad0bf91ccf
• Prompt improvers: built-in tools from OpenAI, Claude that rewrite prompts with best practices
• Give the Agent a Way to Verify Its Work — let agent run tests, not edit them
• Keep sessions short: ~40 messages or one feature per session

AI Tool Directories

• Futurepedia.io: largest AI tools directory
• FutureTools.io
• FindMyAITool: https://findmyaitool.com/
• The Rundown AI: https://supertools.therundown.ai/
• https://aitools.fyi/
• https://aivalley.ai/
• https://theresanaiforthat.com/
• ZDNET: https://www.zdnet.com/article/the-most-popular-ai-tools-of-2024

AI APIs (Free)

• Google Gemini API: multimodal AI, code gen, research agents
  • https://cloud.google.com/vertex-ai/generative-ai/docs/overview
  • Free tier very generous

RAG (Retrieval Augmented Generation)

• All Best RAG Strategies in One MCP: https://www.youtube.com/watch?v=BdZ_GOaaz7M&t=215s

AI Agent Frameworks

• AutoGen (Microsoft): most developer-friendly
• LangGraph (LangChain): visual workflows, fine-grained memory
• CrewAI: structured teams, role-based execution
• MetaGPT: models engineering teams (PM, Dev, QA, Architect)
• AgentVerse: GUI playground for testing agent interactions
• String.com: prompt, run, edit, deploy agents in seconds

LM Arena

• https://lmarena.ai/ — compare model outputs side by side, generous free limits

Models Reference

• Claude: Opus 4.5, Sonnet 4.5, Haiku 4.5
• GPT 5.1
• Gemini 3.0
• GLM-4.6/4.7: best value for coding (~1/7th cost of Claude)
  • Use GLM-4.6 with Cursor: https://medium.com/towards-artificial-intelligence/use-glm-4-6-with-cursor-ee24e3cd1679
• Hugging Face / Qwen 2.5 Coder: https://huggingface.co/collections/Qwen/qwen25-coder-66eaa22e6f99801bf65b0c2f

People to Follow

• “Who to Follow for Claude Code”: https://jpcaparas.medium.com/who-to-follow-if-youre-serious-about-claude-code-0d49abe2d521
• Joe Njenga (prolific CC content)
• Rick Hightower (Spillwave Solutions, skills ecosystem)
• Boris Cherny (CC creator)
• Alireza Rezvani (CC advanced strategies)
• Marco Kotrotsos (skills/agents philosophy)

---

Other

AI Tools (General Purpose)

• Warp AI Terminal: US $18/mn -> $20/mn usage-based
  • https://docs.warp.dev/
  • Warp Drive: context for humans and AI agents
• Gemini CLI: https://geminicli.com/
  • GEMINI.md for project config
  • Cheat sheet: https://medium.com/ai-software-engineer/my-gemini-cli-cheatsheet-will-take-you-from-zero-to-hero-in-minutes-you-need-this-90d4fc1dd6b8
  • Google Cloud Shell Editor: 50 hours/week free
• Google Jules: https://jules.google.com — free (15 tasks/day), supports Astro, uses AGENTS.md
• Google Opal: no-code AI mini-apps, internal automations
• DeepAgent Desktop (Abacus.ai): SOTA coding agent, $10/mn, includes testing
• Comet AI Browser (Perplexity): browser agents with scheduled tasks
• Antigravity (Google IDE): generous limits for Claude Opus/Sonnet: https://kotrotsos.medium.com/how-to-use-google-antigravity-a-complete-guide-3ca56dab0e1f
• AugmentCode.com: large code base specialist — https://medium.com/realworld-ai-use-cases/augmentcode-ai-review-large-code-base-this-is-your-answer-245f7a2cd9a9

Genspark AI

• Build anything: https://www.genspark.ai/
• Demo: https://www.youtube.com/watch?v=ogJg00_U8I8

TaskMaster AI

• Subtask generation, complexity analysis, dependency tracking
• How I reduced 90% errors: https://www.youtube.com/watch?v=1L509JK8p1I
• Works with Cursor and other IDEs
• RooCode Boomerang Tasks (similar)

Private Knowledge / Local AI

• Need local AI models for IP privacy and overnight agentic work
• Doc-Serve Agent Skill: https://github.com/SpillwaveSolutions/doc-serve-skill — private knowledge vault for coding agents
• AnythingLLM: https://anythingllm.com/ — any LLM, AI agents, full local/offline, FOSS
• Local FOSS AI Agentic:
  • Devstral Small 2: cloud-grade coding locally
  • Optimal Setup (~$3,000): AMD Ryzen AI Max 395, 64-128GB RAM, RX 7900 XTX/RTX 4090
  • Browser-Use + CrewAI: max flexibility
  • Nanobrowser: zero-setup browser automation
• Run local models faster with Llamafile (30-500% faster than Ollama on CPU)
• Open Web UI: local FOSS LLM interface
• Local AI agent frameworks review: https://medium.com/@orami98/i-spent-3-weeks-testing-every-local-ai-agent-framework-here-are-the-3-that-actually-work-2025-e166ec461744
• Devstral Small 2: https://medium.com/coding-nexus/devstral-small-2-changes-everything-the-first-cloud-grade-coding-model-you-can-truly-run-on-your-c7aa78ba49fd

Knowledge Base AI

• Haystack: Q&A over your own data using LLMs — PDFs, docs, websites
  • https://github.com/deepset-ai/haystack

Writing Tools

• Type.ai: long-form writing with embedded AI
• EasyGen: LinkedIn posts (https://www.easygen.io/)
• NotebookLM: Google AI research/writing assistant
• Quillbot: improve writing clarity
• Storm: https://ai.gopubby.com/storm-the-best-ai-writing-tool-youve-never-heard-of-f29a6c2e4976

Summarizing Tools

• Bearly.ai: speed up content creation/reading (https://bearly.ai/)
• Humata: AI-powered PDF analysis (https://www.humata.ai/)
• Wiseone: Chrome/Edge extension
• Recall: https://www.getrecall.ai/ — summarize, categorize, review
• YT summarizer: Eightify Chrome extension
• HARPA AI: YT video summarizer — https://harpa.ai/guide/summarize-youtube-videos-to-text
• SciSpace: research papers
• DocuAsk: document Q&A
• PDFGPT.io
• Top 5 summarization tools: https://medium.com/geekculture/top-5-ai-powered-summarization-tools-that-will-transform-your-content-3eeb5efb2293

Image/Video Generation

• Recraft: vector-quality images, PNG/JPG to SVG (https://www.recraft.ai/)
• Leonardo AI: https://leonardo.ai/
• Aragon AI: headshot generator
• Mystic (Freepik): powerful image generation
• Fooocus: offline FOSS, built with Gradio (Python/Svelte), nice customizable UI
• Mystic (Freepik): powerful image generation — https://www.freepik.com/ai
• Microsoft Designer: may be better than Canva
• Infography: infographics from blog posts
• VideoGen
• Clipdrop: professional photos
• Luma Dream Machine: AI video
• Adobe Enhance: studio-quality audio
• Product image tools: https://cdmrkt.medium.com/23-ai-tools-for-product-images-revolutionize-your-e-commerce-6dcabf4483a0
• Product pics from blah to brilliant: https://futurepedia.beehiiv.com/p/take-product-images-from-blah-to-brilliant-04-01-2025

Speech

• ElevenLabs: text-to-speech in own voice
• Adobe Enhance: podcast.adobe.com/enhance
• Hedra: photo to video with audio sync

Design (UI/UX)

• Galileo AI: text to UI designs
• Khroma: AI-powered color palettes
• Uizard: wireframes in minutes
• Visily: for non-designers
• VisualEyes: predict user viewing patterns

Presentations

• Gamma.app

Explore / Evaluate Later

• SourceAI: code generation — https://sourceai.dev/ (joined waitlist 2022NO15)
• Roomai.com: redesign rooms with AI
• Flowith: create AI agents — https://flowith.io
• GoOver: deep research — https://intro.goover.ai
• Promptize app: https://app.promptize.net/
• NotebookLM podcasts: https://www.zdnet.com/article/googles-hidden-ai-tool-turns-your-text-into-stunningly-lifelike-podcasts-for-free-listen-for-yourself
• Create AI chat: clone personality, expertise

Business Tools

• CheckMyIdea: https://www.checkmyidea-ia.com/ — evaluate business idea potential #use
• GoReply: https://goreply.com/ — online consultancy sessions, model for Sherpa business #eval
• ReclaimAI: calendar management
• Pickaxe: no-code AI products
• GravityWrite: website/social media content
• Google Labs: labs.google.com
• Forms.app: professional forms/surveys

Free AI Tools

• Riverside: free audio/video transcription (https://riverside.fm/transcription)
• Napkin AI: text to visuals
• Tango: interactive workflow guides
• GigaBrain: AI for Reddit + Stack Overflow
• TLDRthis: auto-summarize documents

General GPTs

• 10 best ChatGPT alternatives: https://medium.com/@affineworkos/10-best-chatgpt-alternatives-you-can-look-for-cb77c3eac5cf
• Notdiamond.ai: routes to best LLM (API key: sk-b67…951) — use Arena mode
• Perplexity: chat with sources
  • Alternative: Komo (https://komo.ai/)
• ChatGPT: https://chatgpt.com (plugins listed: Portfolio Pilot, Webpilot, etc.)
  • ChatGPT Chrome Extensions: YouTube Summary, ReaderGPT, YoutubeDigest
    • https://beingpax.medium.com/10-awesome-chatgpt-extensions-to-improve-your-life-590a530ddde6
  • Custom GPTs for n8n workflows: https://medium.com/@milhoornaert/stop-using-normal-chatgpt-here-are-3-custom-gpts-to-use-instead-41ed407d7572
• Heypi.com
• You.com: summarize URLs, YT, files
• Poe.com

FOSS Alternatives

• ElevenLabs -> Speechify
• Canva -> Microsoft Designer
• Veed.io -> CapCut (free, powerful video editing)
• HeyGen -> Wondershare Virbo

n8n Automation

• Self-hosted AI starter kit: https://github.com/n8n-io/self-hosted-ai-starter-kit
• Custom GPTs for n8n workflows #eval

AI Browser Tabs (Archived/Reference)

• https://aistudio.google.com
• https://gemini.google.com
• https://chatgpt.com
• https://claude.ai
• https://www.perplexity.ai
• https://poe.com

Subscriptions Summary

• Warp: ~$20/mn usage-based
• Cursor: US $20/mn
• Claude Code: billing 29th

25 Prompts to Teach AI Your Thinking Style

• https://medium.com/write-a-catalyst/25-prompts-to-teach-chatgpt-5-your-thinking-style-a757611dc361

6-Step PEST Framework for Strategic Decisions

• https://medium.com/publishous/the-6-step-framework-for-mastering-strategic-decisions-using-ai-and-pest-7fd76224dcfa

---

Uncategorized Other Items

Items recovered from original AI_from_Dynalist.md that were not included in the sorted categories above. Organized by type.

Action Items / Tasks

• Produce executive summary of “The AI Agent Race is Over” article for compliance, privacy, and operations manager role at Lifemark. Email to Jim Wernham@lifemark.ca
• INCUBATE ON: Skills concept — ongoing incubation for how skills apply to solo dev, SMART DEBT coach, marketing, content
• Research Claude skills repositories and use some, learn from some. Check Anthropic’s cookbook or community repos for examples
• Save Mastering Agentic Skills (Spillwave) article to Clippings
• Explore Spillwave resources — “Great tips, and many useful resources to explore”
• Continue reading CLAUDE.md Guide article at “Building Your Agent Army: Specialized Agents”
• Use NotebookLM to summarize CC cheatsheet for my needs: Astro monorepo, SD Mission, KB, utils, SD App
• Learn about CC 2.3
• Learn about “best” AI Agents and best frameworks to create them (for my needs): web dev, strategic/dev/ops planning
• “Get prompt” from “I accidentally made Claude 45% smarter” article
• Explore MCP servers for Cursor as well (Cline has MCP Marketplace)
• Learn how to use Projects inside AI chats (Gemini, Claude, ChatGPT) for different development contexts

Workflow Patterns / Prompts

• Skill capture prompt: “Create a Skill that captures the pattern we just used for [task]. Focus on [specific aspects].”
• Iterative Upgrade “Creator” Loop: Feed results back to skill-creator: “The output was too verbose; update the skill to be more concise.” It rewrites SKILL.md for you.
• Automated Evaluation Test Harness: Create a script that runs your skill against 5 different inputs and saves output. Ask Claude to “evaluate these 5 outputs against my SKILL.md guidelines and grade them.”
• TASKS.md 24/7 automation prompt: “Please pull the next task from TASKS.md and fully implement and test it. After running tests, analyze logs and screenshots to verify correctness. If you spot any failures or issues, fix them and rerun tests — keep iterating until everything passes. Once verified, commit your changes, then fetch the next task from TASKS.md.”
• Multi-agent workflow pattern: Model the high-level workflow using prompt.md to create specs, plan, tasks
• Always include in prompts: “ask me any clarifying questions needed to ensure you clearly understand the task”
• AI coding process: Include step to analyze carefully for any security vulnerabilities
• Store prompts in a .prompts folder for organization
• Prompt workflow: Review first draft of prompt and improve first. Iterate on results.
• AI dev step: Get AI to consider overlooked edge cases
• Core Claude workflow: Give Claude context, constraints, and objectives. Let it work. Review outcome and adjust documentation.
• CC in Cursor question: Best to use CC as VS Code extension — why not in Cursor as extension? #investigate

AI Dev Tactics (Missing from Main Sections)

• Use popular languages, frameworks (better AI training data)
• Use a starter project
• Use one repository (front and backend)
• Leverage Cursor Composer: apply changes to multiple files
• Context: use @ to reference several files in Cursor
• Commit often
• Understand: have AI explain code you don’t understand

Missing URLs / Resources

• https://claude.com/product/claude-code — official CC product page
• D:\FSS\KB\Business\03_Processes\AI\04_Tools & Platforms\Claude — local KB cross-reference
• https://www.warp.dev/blog — Warp Blog
• https://www.aragon.ai/ — AI Headshot Generator
• https://www.tango.us/ — Tango: auto-generates interactive workflow guides
• https://www.tldrthis.com/ — TLDRthis: auto-summarize documents
• https://www.microsoft.com/en-in/microsoft-copilot/ — Microsoft Copilot
• https://github.com/modelcontextprotocol/servers — main MCP servers repo (root)
• https://youtube.com/watch?v=LttOhQzE3kE&si=2T2uVGW67-zKuRfl — “I Built an MCP Server in 20 Minutes Using Only AI”
• https://console.anthropic.com/dashboard — Anthropic console (prompt generator)
• https://skillzwave.ai/ — world’s largest agentic skill marketplace (Rick Hightower / Spillwave)
• Rick Moretti YouTube: “five mind-blowing use cases of Claude skills” (~November 13th) — create strategic decisions skill

Missing MCP Servers / Concepts

• Browserbase MCP Server — cloud browser automation
• Svelte MCP Server — placeholder, explore options
• Cloudflare MCP Server — and other hosting platforms (Netlify, Vercel, GCP)
• MCP Server creation tutorial: “I Built an MCP Server in 20 Minutes Using Only AI” (link above)
• Sequential Thinking MCP detailed features: break down complex problems into manageable steps, revise/refine thoughts as understanding deepens, branch into alternative reasoning paths, adjust total number of thoughts dynamically, generate and verify solution hypotheses

Missing ChatGPT Plugins (Full List)

• Portfolio Pilot, Public, Buywisely, Charwithvideo, Webpilot, Browse with Bing, AI Tool Hunt
• (Bold = missing from sorted file’s abbreviated list)

Missing Tool

• Mixo: create website from prompt (found under FutureTools.io in original)