Arindam Majumder 𝕏
@Arindam_1729 • 9,316 subscribers
a builder. founder @studio1hq youtube: https://t.co/ZE1x8oT6Ac
Shorts
Videos
Claude Code has 3 models. Stop making Sonnet do Haiku’s job. Sonnet 5.5 builds. Haiku 5.5 scouts. Opus 5.5 reviews. Put them in parallel Git worktrees and let each model do what it’s best at. The setup: → Sonnet 5.5: Build features across isolated worktrees, writing code and running tests without touching your main branch. → Haiku 5.5: Explore the repo, index the AST, find dependencies, and scout for implementation paths while Sonnet codes. → Opus 5.5: Review architecture before implementation, investigate repeated test failures, and verify diffs before merging. Launch separate sessions: claude --worktree feat-auth claude -w feat-scout --model haiku Each session gets its own working directory and Git branch. Your agents can edit files and run tests in parallel without overwriting each other’s work. The full setup goes further: .worktreeinclude shares local configs across worktrees. Auto-cleanup keeps finished sessions tidy. Supervisor checkpoints catch contract drift and repeated test failures. A cwt alias makes launching new worktrees one command. Opus reviews the final diff before anything reaches main. The goal is simple: stop wasting expensive model calls on repo exploration, stop serializing work that can run in parallel, and stop merging code without a proper review. Paste the setup prompt into Claude Code. Ask it to show every config diff first, then approve the changes before anything gets edited. Build with Sonnet. Scout with Haiku. Review with Opus. Ship in parallel.
Arindam Majumder 𝕏14,192 次观看 • 11 天前
Just tested Kimi.ai K3 against GPT-5.6 and Grok 4.5 on the same task: build a Subway Surfers game. Kimi K3 was my favorite. Better UI, smoother gameplay, and the overall experience felt more polished. GPT-5.6 was solid too, while Grok 4.5 fell behind on the UI. The interesting part is that all three needed at least one follow-up prompt. Kimi also took a bit longer to generate, but the extra time paid off. Ran Kimi K3 through Nebius Token Factory. Low inference cost + a really capable open source model is a great combo.
Arindam Majumder 𝕏46,107 次观看 • 2 个月前
The MCP finally has proper authentication! It now supports OAuth 2.1 out of the box. That means proper authentication for AI agents and MCP tools is finally standardized. No more hacked-together API keys or custom token logic. With OAuth 2.1 + MCP, you can now: - Dynamically register clients - Issue & verify scoped JWT tokens - Track which agent acted for which user - Plug into existing IdPs (Auth0, Clerk, Firebase, etc.) I’ve been testing it with Scalekit, a lightweight OAuth layer built for MCP servers. It adds full discovery, token flow, and audit logs with minimal config. This is the missing piece for secure, multi-agent systems. Auth that actually works.
Arindam Majumder 𝕏11,525 次观看 • 1 年前
没有更多内容可加载