Evals for Claude agents: measure quality, gate releases
Build an eval loop for Claude agents: define quality, pick graders, use LLM judges well, and gate every release on automated checks, not vibes.
8 min read
Agentic AI & LLMs
Agentic AI & LLMs archive page 44 of 631
Agentic AI & LLMs
Agentic AI, LLM engineering, and the models behind modern automation — multi-agent systems, LLM evaluation and comparisons, RAG, fine-tuning, AI infrastructure, security, and production AI engineering.
From the blog
Latest articles
Securing Claude agents: sandboxing & prompt injection
Harden Claude agents: sandbox tool execution, enforce least privilege, protect secrets, and defend against prompt injection from untrusted content.
8 min read
Cutting Claude agent token cost: caching & batching
Keep Claude agent runs cheap and fast: prompt caching, request batching, model routing across Opus/Sonnet/Haiku, and trimming bloated context.
7 min read
Debugging Claude agents: loops, bad tool calls, fixes (Rethinking How We Build)
Why Claude agents loop, call the wrong tool, or invent arguments — and the trace logging, loop guards, and tool contracts that fix it fast.
8 min read
Context engineering for Claude: what to include and cut
Design context for Claude agents in 2026: what belongs in the window, what to leave out, just-in-time retrieval, and compaction for long, reliable runs.
7 min read
Wiring MCP tools into Claude: auth, schemas, and errors
Production MCP integration for Claude agents in 2026: scoped auth, strict schemas, structured error handling, and idempotency for safe, reliable tool calls.
7 min read
Claude agent design patterns: tools, prompts, and context
Reusable 2026 patterns for reliable Claude agents: narrow tools, layered prompts, just-in-time context, subagent isolation, skills, and boundary validation.
7 min read
Build your first Claude agent: a step-by-step walkthrough
A runnable 2026 walkthrough for building a Claude agent: agent loop, tool definitions, context, error handling, and graduating to MCP and the Agent SDK.
7 min read
Inside Claude Code's architecture: how the agent loop works
How Claude Code works internally in 2026: the agent loop, 1M-token context, parallel subagents, MCP, skills, and hooks — and how they fit together end to end.
8 min read