[{"data":1,"prerenderedAt":45},["ShallowReactive",2],{"$fWhdZ67ic7EGZ7gSqJvuFjkjeYiLeKt29V3-mvODS5wc":3},{"date":4,"generated_at":5,"picks":6,"candidates_scanned":43,"candidates_scored":44},"2026-03-14","2026-03-14T05:49:19.674874+00:00",[7,21,32],{"rank":8,"title":9,"source":10,"url":11,"category":12,"tldr":13,"score":14,"scores":15,"why":20},1,"[Release] anthropics/claude-agent-sdk-typescript: v0.2.76","GitHub anthropics/claude-agent-sdk-typescript","https://github.com/anthropics/claude-agent-sdk-typescript/releases/tag/v0.2.76","Release","- You can now branch off conversations from a specific point using `forkSession()`, which lets you explore different paths without losing your original thread\n\n- There's a new way to cancel queued messages before they execute by using `cancel_async_message`, so you can clean up the backlog if needed\n\n- Developers working with the ExitPlanMode tool now get a `planFilePath` field they can use to access plan files programmatically\n\n- The SDK now supports MCP (Model Context Protocol) server requests more smoothly—it can handle when those servers ask for input and send back completion messages automatically\n\n- This version syncs up with Claude Code v2.1.76, so you're getting all the same improvements from that release",89,{"direct_claude_relevance":16,"practical_utility":17,"novelty":18,"source_credibility":19},34,28,17,15,"Official SDK release with significant new features (forkSession, cancel_async_message, MCP elicitation hooks) and direct update instructions. Developers using Claude Agent SDK need this immediately to access new branching and control capabilities.",{"rank":22,"title":23,"source":24,"url":25,"category":12,"tldr":26,"score":27,"scores":28,"why":31},2,"[Release] anthropics/claude-code-action: v1.0.72","GitHub anthropics/claude-code-action","https://github.com/anthropics/claude-code-action/releases/tag/v1.0.72","- They made the tag mode tool more secure by protecting it against prompt injection attacks, which is when someone tries to trick the system by sneaking extra instructions into their prompts\n- This is a small security patch to make sure bad actors can't manipulate how the tool works by being clever with their input",87,{"direct_claude_relevance":16,"practical_utility":29,"novelty":30,"source_credibility":19},20,16,"Official Anthropic release of Claude Code Action with security hardening against prompt injection. While modest in scope, official releases from anthropics GitHub repos warrant high scoring as power users need to stay current on dependencies.",{"rank":33,"title":34,"source":35,"url":36,"category":12,"tldr":37,"score":38,"scores":39,"why":42},3,"v2.1.76","Claude Code Releases","https://github.com/anthropics/claude-code/releases/tag/v2.1.76","- MCP servers can now ask for information mid-task through interactive dialogs, and there are new hooks to intercept and customize responses before sending them back\n\n- A bunch of bug fixes covering everything from voice mode issues on Windows to a circuit breaker preventing auto-compaction from retrying forever, plus improvements to how permissions work and how the app handles remote connections",82,{"direct_claude_relevance":40,"practical_utility":17,"novelty":41,"source_credibility":19},35,14,"Official Claude Code release with substantial new features and fixes. MCP elicitation support, sparse checkout for monorepos, slash commands, and numerous bug fixes directly impact developer workflows. Highly actionable for daily Claude Code users.",19,12,1776402243522]