Topic

Anthropic Agent SDK

Also known as: Claude Agent SDK, Anthropic Claude Agent SDK, Claude Code Hooks SDK

The programmatic TypeScript surface for building agents on top of Claude. Covers the 12 hook events (PreToolUse, PostToolUse, UserPromptSubmit, Stop, SubagentStop, PreCompact, SessionStart, SessionEnd, Notification, PermissionRequest, SubagentStart, PostToolUseFailure), the HookCallback signature, hookSpecificOutput decisions (allow/deny/ask + updatedInput), and how programmatic hooks compare to declarative Claude Code settings.json hooks.

Canonical resource

/blogs/anthropic-agent-sdk-hooks-typescript-reference

Articles (3)

Anthropic Agent SDK Hooks: TypeScript Reference

Programmatic Anthropic Agent SDK hooks reference. The core events, full TypeScript signatures, PreToolUse, PostToolUse, Stop, SubagentStop, UserPromptSubmit with worked examples.

May 23, 2026

Primary

Claude Code Hooks: Corrections to Our February 2026 Guide (Dated Note)

A dated note on our February 2026 Claude Code hooks guide: why its config examples would not have worked, the correct shape from the official reference, and what we run instead.

Feb 14, 2026

Primary

Context Engineering Beyond CLAUDE.md: The 5-Layer Hierarchy

CLAUDE.md is just layer one. The five-layer context hierarchy, memory patterns, and subagent strategies that separate productive AI coding from prompt guessing. With working examples.

Feb 14, 2026

Articles (3)

Related Topics