AI DEVELOPMENT FIELD NOTES
Practical AI development, organized through primary sources and hands-on verification¶
From AI development tools to related open-source software, new capabilities, and model updates, SmartScope helps practitioners evaluate, test, and use what matters.
Instead of repeating announcements or specifications, we use primary sources and hands-on tests to explain procedures, comparisons, underlying mechanisms, and practical impact.
- Primary sources
- Hands-on tests
- Practical judgment
FIND YOUR ANSWER
Find an article by task or symptom¶
Filter by what you want to do, the symptom you are seeing, or the article's primary genre. Tags preserve cross-cutting topics.
Tool
Genre
| Tool | What you want to do or fix | Article | Genre |
|---|---|---|---|
| Claude | Let Claude sign in to websites without exposing passwords | 1Password for Claude Explained | Latest & models |
| Claude | Delegate multi-step computer work to Claude without a terminal | Using Claude Cowork | Operations & automation |
| Claude | Connect work data from Notion, Slack, and other services to Claude | Claude Connectors Directory | Operations & automation |
| Claude | Connect an internal tool or custom API to Claude safely | Custom MCP for Claude | Development platform |
| Claude | Fix a Skill that does not trigger reliably | Fix Skill Triggering | Troubleshooting |
| Claude | Audit and revoke shared Claude chats and published artifacts | 🔐 Audit Claude Share Links | Latest & models |
| Claude | Understand how Claude Skills work and how to install them | What Are Claude Agent Skills | OSS & primary sources |
| Claude | Choose between Skills and Projects for a Claude workflow | ⚖️ Skills vs Projects | OSS & primary sources |
| Claude | Design a reliable, reusable Claude Skill | Claude Skills Design Patterns | Operations & automation |
| Claude | Evaluate and improve Skill triggering and output quality | skill-creator Evaluation | Operations & automation |
| Claude | Build a first agent with the Claude Agent SDK | Claude Agent SDK Beginner Guide | Development platform |
| Claude | Resume and monitor a long-running Claude Agent SDK task safely | Long-Running Agent SDK | Operations & automation |
| Claude | Decide whether to take the Claude Certified Architect exam | Claude Certified Architect | Enterprise & teams |
| Claude | Decide how to handle watermarks and disclosure for text edited or translated with Claude | Claude AI Watermarks | Latest & models |
| Claude Code | Reduce approval prompts for automated work | 🔐 Auto-Permission Guide | Operations & automation |
| Claude Code | Install Claude Code correctly for my operating system | 🚀 Installation | Development platform |
| Claude Code | Diagnose and clean up configuration debt | Claude Code /checkup | Latest & models |
| Claude Code | Fix API Error 401 when /login also fails | Fix 401 Auth Error | Troubleshooting |
| Claude Code | Coordinate independent Claude Code sessions with safe permission boundaries | Claude Code Cross-Session Messaging | Enterprise & teams |
| Claude Code | Fix a claude command that is not recognized | 🔧 Not Recognized Error Fix | Troubleshooting |
| Claude Code | Share screenshots or images with Claude Code | 🖼️ Image Attachment Guide | Operations & automation |
| Claude Code | Compare compact behavior with Codex | Claude Code vs Codex Compaction | OSS & primary sources |
| Codex | Finalize a plan before implementation | 🎯 Codex Plan Mode Guide | Operations & automation |
| Codex | Reduce approval prompts safely | ✅ No-Approval Execution Guide | Operations & automation |
| Codex | Allow npm, curl, and API calls without removing the filesystem sandbox | 🌐 Network Restriction Fix | Troubleshooting |
| Codex | Recover missing sidebar history after moving Codex from WSL to Windows | 🧭 Recover Codex History | Troubleshooting |
| Codex | Decide when an automation must stop | Codex Stop Conditions | Operations & automation |
| Codex | Fix Codex MCP EISDIR errors on WSL | 🐧 Codex WSL Performance | Troubleshooting |
| Codex | Diagnose Codex Automation failures | Automation Health Check | Troubleshooting |
| Codex | Fix a Codex CLI update that still launches an older binary | Codex CLI Update | Troubleshooting |
| Codex | Scan a repository for vulnerabilities with Codex and add it to CI | Codex Security CLI in CI | OSS & primary sources |
| Codex | Recover after reaching the context limit | Context Window Overflow Error Fix | Troubleshooting |
| Codex | Fix missing field sandboxPolicy | 🧩 sandboxPolicy Error | Troubleshooting |
| Codex | Use full access safely inside an externally isolated environment | Full-Access Safety | Operations & automation |
| GitHub Copilot | Design custom instructions | 📖 Custom Instructions | Operations & automation |
| GitHub Copilot | Write Agent Skills with SKILL.md | 🧩 Agent Skills | Operations & automation |
| GitHub Copilot | Reuse shared instructions across AI coding tools | 📋 AGENTS.md Shared Instructions | Operations & automation |
| GitHub Copilot | Apply different Copilot instructions by file type | 🎯 applyTo Patterns | Operations & automation |
| GitHub Copilot | Delegate an issue to Copilot and get reviewable changes | 🤖 Copilot Cloud Agent | Operations & automation |
| GitHub Copilot | Understand usage and billing | Copilot AI Credits Billing | Enterprise & teams |
| GitHub Copilot | Choose GitHub Copilot models by workload without wasting AI Credits | Copilot Model Value | Latest & models |
| Methods & Organizations | Inspect how an agent loop behaves | Loop Anatomy | OSS & primary sources |
| Methods & Organizations | Decide how much harness engineering to build | Harness Engineering — What to Build in Practice | OSS & primary sources |
| Methods & Organizations | Choose between Claude Code and Codex and route development work | Claude Code vs Codex | Latest & models |
| Methods & Organizations | Choose between GPT-5.6 Sol and Claude Fable 5 | Sol vs Fable 5 | Latest & models |
| Methods & Organizations | Turn individual AI gains into organizational results | AI Adoption Stages | Enterprise & teams |
| Methods & Organizations | Define how a team should route work across Chat, Work, and Codex | Chat and Work Today | Latest & models |
| Methods & Organizations | Decide whether Seedance 2.5 is ready for a current video production workflow | Seedance 2.5 Spec Gaps | Latest & models |
| Methods & Organizations | Assign ownership for exceptions before an enterprise rollout | AI Exception Ownership | Enterprise & teams |
| Methods & Organizations | Review GitHub permissions and development-environment controls | Money Forward GitHub Breach | Enterprise & teams |
| Methods & Organizations | Decide whether Muse Code is worth testing as an alternative to Claude Code or Codex | Muse Code Performance and Price | Latest & models |
| Methods & Organizations | Control company-wide AI spend while allocating premium models to the work that needs them | Allocate Enterprise AI with FinOps | Enterprise & teams |
| Methods & Organizations | Create Excel, PowerPoint, and Word files with Claude and review them reliably | 📊 Claude Document Skills | Operations & automation |
| Methods & Organizations | Prevent a shared AI credit pool from running out mid-cycle | AI Credit Pool Scheduling | Enterprise & teams |
| Methods & Organizations | Decide whether more instructions actually improve LLM performance | Separate AI Freedom from Measurement | OSS & primary sources |
| Methods & Organizations | Decide which Codex tasks should use Fast mode | Choosing Codex /fast | Latest & models |
| Development platform | Run GitHub Actions steps concurrently without splitting jobs | Actions Step Parallelism | Development platform |
| Development platform | Choose between GitHub Pages and Cloudflare | Free Hosting Choice | Development platform |
| Development platform | Give Claude Code or Codex semantic codebase understanding | Serena MCP | Development platform |
| Development platform | Keep development and deployment working during a GitHub outage | GitHub Outage Resilience | Development platform |
| Development platform | Unify multiple LLM APIs behind one endpoint with budget and key controls | What Is LiteLLM? | OSS & primary sources |
| Development platform | Create and validate Office files from the command line | What Is OfficeCLI | OSS & primary sources |
| Development platform | Run Hugging Face's voice agent with a fully local stack | Open-Source Voice Agent | OSS & primary sources |
| Development platform | Convert PDFs with tables, diagrams, or formulas into LLM-ready Markdown | What Is MarkPDFDown? | OSS & primary sources |
| Development platform | Decide whether Cloudflare OS is ready for a company-wide AI agent rollout | Cloudflare OS Adoption | Enterprise & teams |
No matching article. Try the tag index or the site search in the header.
RECENTLY PUBLISHED
Latest articles¶
New tests, open-source explainers, and the practical impact of recent updates, ordered by publication date.
- GitHub Outages in 2026: A Resilience Plan and the State of Forge Alternatives
- Was SpaceX's Cursor Acquisition a Side Note? How Origin Could Loosen GitHub's Grip
- Claude Code vs Codex: Which Is Better in August 2026?
- Choosing GitHub Copilot Models in 2026: Routing Eight Models by Effective Cost
- What Claude's AI Watermark Means: Article 50, Machine-Readable Marks, and Their Limits
- Inside Cloudflare OS: What to Decide Before Giving Every Employee an AI Agent
- Can Seedance 2.5 Really Deliver 4K and 180 Seconds? Reading the Official Spec Gaps
- Claude Code Cross-Session Messaging: Permission Boundaries and Organization Controls
EDITORIAL STANDARD
What SmartScope covers¶
In a fast-moving field, clarity comes not only from adding information, but also from deciding what not to duplicate.
Publish what can be verified
Hands-on tests of AI development tools, explanations of relevant open-source architecture, and practical analyses of new capabilities and model updates grounded in primary sources.
Do not simply reproduce catalogs
Specifications, pricing, and model lists point to official sources when they change quickly. SmartScope adds context, comparison, verification, and practical use cases.
Update alongside new evidence
New tests, open-source explainers, and release coverage refresh the latest feed, topic paths, and the article index whenever useful.
EXPLORE THE LIBRARY
Browse all topics¶
For subjects beyond the reverse lookup, use the continuously maintained pillar pages.
