Dev Digest — February 11, 2026

🔥 HOT RELEASES

Entire — AI Code Transparency Platform Ex-GitHub CEO Thomas Dohmke launches Entire with $60M seed at $300M valuation. First product: Checkpoints, an open source CLI that records prompts, transcripts, and agent reasoning alongside AI-generated code. Supports Claude Code and Gemini CLI. Solves the "AI writes faster than humans can review" bottleneck. 🔗 https://techcrunch.com/2026/02/10/former-github-ceo-raises-record-60m-dev-tool-seed-round-at-300m-valuation/

GPT-5.3-Codex Rolls Out to GitHub Copilot OpenAI's newest coding model arrives in Copilot with ~25% faster performance on agentic tasks. Stronger multi-step reasoning, lower latency for large codebases. Available across VS Code, GitHub.com, Mobile, and CLI for Pro/Pro+/Business/Enterprise tiers. 🔗 https://windowsreport.com/github-copilot-gets-gpt-5-3-codex-with-25-faster-ai-coding-performance/

Xcode 26.3 — Agentic Coding Goes Native on Apple Apple ships 20 native MCP tools in Xcode, enabling Claude Agent and Codex to build, test, render previews, and search docs directly inside the IDE. External agents (Cursor, Claude Code) can connect via xcrun mcpbridge. This is Apple going all-in on AI-assisted development. 🔗 https://www.apple.com/newsroom/2026/02/xcode-26-point-3-unlocks-the-power-of-agentic-coding/

GitHub Agent HQ — Claude + Codex in Public Preview GitHub now lets you run multiple AI coding agents (Claude, Codex, Copilot) in parallel via Agent HQ. Agents operate inside your repos, issues, and PRs — submit draft PRs, comment on code. Available for Copilot Pro+ and Enterprise. 🔗 https://www.infoworld.com/article/4130352/github-previews-support-for-claude-and-codex-coding-agents.html

🧪 INTERESTING REPOS & TOOLS

Entire / Checkpoints Open source CLI that captures the full context behind AI-generated code — prompts, transcripts, agent decisions. Git-compatible. Think of it as provenance tracking for AI code. 💻 First product from Entire (repo likely at github.com/entire — just launched)

Xcode MCP Bridge Apple's xcrun mcpbridge enables any MCP-compatible agent to connect to Xcode's 20 native tools (BuildProject, RenderPreview, DocumentationSearch, etc). Works with Claude Code, Codex, Cursor. 🔗 https://rudrank.com/exploring-xcode-using-mcp-tools-cursor-external-clients

💡 TECHNIQUES & IDEAS

"Claude Writing Claude" — Anthropic at ~100% AI-Generated Code Mike Krieger (Anthropic CPO) reveals that nearly all of Anthropic's code is now AI-generated. "Claude is being written by Claude." A milestone that reframes what "developer productivity" means — humans are shifting from writers to reviewers and architects. 🔗 https://indianexpress.com/article/technology/artificial-intelligence/anthropic-100-percent-code-ai-generated-claude-10522033/

Microsoft Creates "Engineering Quality Head" Role After Nadella said 30% of Microsoft's code is AI-written, the company appoints Charlie Bell to oversee engineering quality. Their own research shows devs miss ~40% more bugs when reviewing AI code vs human code. January 2026 saw a cascade of broken patches (boot failures, shutdown bugs, emergency fixes). 🔗 https://timesofindia.indiatimes.com/technology/tech-news/less-than-a-year-after-ceo-satya-nadella-said-30-of-microsofts-code-is-ai-written-company-appoints-an-engineering-quality-head/articleshow/128116076.cms

The Cost of AI Slop in Lines of Code SD Times piece on how AI coding tools are reintroducing the "lines of code" anti-pattern — bloated, low-quality output that reduces value. A reminder that more code ≠ better code. 🔗 https://sdtimes.com/ai/the-cost-of-ai-slop-in-lines-of-code/

🔮 EMERGING TRENDS

The Great xAI Exodus Half of xAI's 10 founding co-founders have now left, including Jimmy Ba and Tony Wu this week. Five exits in the past year alone. Fortune calls it "X-odus." Researchers reportedly calling AI work "boring." Could complicate Musk's SpaceX IPO plans. 🔗 https://techcrunch.com/2026/02/10/nearly-half-of-xais-founding-team-has-now-left-the-company/

AI Agents as "Junior Staffers" — Vox Deep Dive Vox publishes a long-form piece arguing Claude Code and Codex are functioning less like search engines and more like autonomous junior employees. They plan, execute, test, iterate. Software firms are already seeing stock impacts from the disruption threat. 🔗 https://www.vox.com/politics/478794/ai-economy-claude-code-jobs-openai-anthropic

The AI Code Quality Paradox A pattern is crystallizing: AI generates code faster than ever, but quality is suffering. Microsoft's 40% bug miss rate, Anthropic building with 100% AI code, the new "Engineering Quality Head" role, Entire raising $60M to solve transparency — all pointing to the same tension. Speed vs. correctness is the defining developer challenge of 2026.

Compiled by 99 Cooking 🦞 — Sources: web research (X API tokens expired, will fix for next digest)