Central authority for Claude Agent SDK (TypeScript and Python SDKs). Covers SDK installation, authentication (Anthropic key, Bedrock, Vertex), sessions and resumption, forking sessions, streaming vs single mode, custom tools, permissions (allowedTools, disallowedTools, permissionMode), MCP integration, system prompts (CLAUDE.md, appendSystemPrompt, outputStyle), cost tracking, todo tracking, structured outputs, hosting patterns, plugins, and SDK branding guidelines. Assists with building custom agents, configuring SDK options, and troubleshooting SDK issues. Delegates 100% to docs-management skill for official documentation.
Provides guidance for Claude Agent SDK development and configuration.
/plugin marketplace add melodic-software/claude-code-plugins/plugin install claude-code-observability@melodic-softwareThis skill is limited to using the following tools:
references/testing-and-notes.mdSTOP - Before providing ANY response about Claude Agent SDK:
- INVOKE
docs-managementskill- QUERY using keywords: Agent SDK, TypeScript SDK, Python SDK, sessions, custom tools, permissions, or related topics
- BASE all responses EXCLUSIVELY on official documentation loaded
Skipping this step results in outdated or incorrect information.
Before responding, verify:
If ANY checkbox is unchecked, STOP and invoke docs-management first.
Central authority for Claude Agent SDK (TypeScript and Python SDKs). This skill uses 100% delegation to docs-management - it contains NO duplicated official documentation.
Architecture: Pure delegation with keyword registry. All official documentation is accessed via docs-management skill queries.
This skill follows a pure delegation architecture with three loading layers:
All official Agent SDK documentation is accessed through docs-management queries:
| doc_id | Topic |
|---|---|
| platform-claude-com-docs-en-agent-sdk-overview | SDK overview and getting started |
| platform-claude-com-docs-en-agent-sdk-typescript | TypeScript SDK API reference |
| platform-claude-com-docs-en-agent-sdk-python | Python SDK API reference |
| platform-claude-com-docs-en-agent-sdk-sessions | Session management and resumption |
| platform-claude-com-docs-en-agent-sdk-streaming-vs-single-mode | Input mode selection |
| platform-claude-com-docs-en-agent-sdk-custom-tools | Creating custom tools |
| platform-claude-com-docs-en-agent-sdk-permissions | Permission configuration |
| platform-claude-com-docs-en-agent-sdk-modifying-system-prompts | System prompt customization |
| platform-claude-com-docs-en-agent-sdk-cost-tracking | Cost and usage tracking |
| platform-claude-com-docs-en-agent-sdk-todo-tracking | Todo list integration |
| platform-claude-com-docs-en-agent-sdk-structured-outputs | Structured output features |
| platform-claude-com-docs-en-agent-sdk-hosting | Production hosting patterns |
| platform-claude-com-docs-en-agent-sdk-mcp | MCP server integration |
| platform-claude-com-docs-en-agent-sdk-plugins | Plugin loading |
| platform-claude-com-docs-en-agent-sdk-subagents | Subagent usage |
| platform-claude-com-docs-en-agent-sdk-skills | Skills integration |
| platform-claude-com-docs-en-agent-sdk-slash-commands | Slash command usage |
This skill has no local reference files. All documentation is accessed via docs-management queries using the keyword registry below. This ensures:
Keywords: Agent SDK, Claude Agent SDK, TypeScript SDK, Python SDK, npm install claude-agent-sdk, pip install claude-agent-sdk, sessions, session resumption, forking sessions, streaming mode, single mode, custom tools, allowedTools, disallowedTools, permissionMode, system prompts, appendSystemPrompt, outputStyle, CLAUDE.md, settingSources, cost tracking, todo tracking, structured outputs, hosting, plugins SDK, MCP SDK, agent branding
Use this skill when:
Use these keywords when querying docs-management skill for official documentation:
| Topic | Keywords |
|---|---|
| Overview | "Agent SDK", "Claude Agent SDK overview", "why use Agent SDK" |
| Installation | "Agent SDK installation", "npm install claude-agent-sdk", "pip install claude-agent-sdk" |
| SDK Options | "SDK options", "TypeScript SDK", "Python SDK" |
| Migration | "Claude Code SDK migration", "migration guide" |
| Topic | Keywords |
|---|---|
| API Key | "Agent SDK authentication", "ANTHROPIC_API_KEY" |
| Bedrock | "CLAUDE_CODE_USE_BEDROCK", "Bedrock SDK" |
| Vertex | "CLAUDE_CODE_USE_VERTEX", "Vertex SDK" |
| Topic | Keywords |
|---|---|
| Session Management | "session management SDK", "session ID", "getting session ID" |
| Resumption | "session resumption", "resume option", "continuing sessions" |
| Forking | "forking sessions", "forkSession", "session branching" |
| Topic | Keywords |
|---|---|
| Overview | "streaming vs single mode", "input modes SDK" |
| Streaming Mode | "streaming input mode", "recommended mode" |
| Single Mode | "single input mode", "non-streaming" |
| Topic | Keywords |
|---|---|
| Creating Tools | "custom tools SDK", "creating custom tools" |
| Tool Schema | "tool schema", "tool definition SDK" |
| Tool Handlers | "tool handlers", "tool execution SDK" |
| Topic | Keywords |
|---|---|
| Tool Permissions | "allowedTools", "disallowedTools", "tool permissions SDK" |
| Permission Mode | "permissionMode SDK", "permission strategy" |
| Handling Permissions | "handling permissions SDK", "permission callbacks" |
| Topic | Keywords |
|---|---|
| Overview | "modifying system prompts", "system prompts SDK" |
| CLAUDE.md | "CLAUDE.md SDK", "settingSources", "project instructions" |
| Append System Prompt | "appendSystemPrompt", "custom instructions" |
| Output Style | "outputStyle SDK", "agent persona" |
| Topic | Keywords |
|---|---|
| Cost Tracking | "cost tracking SDK", "token usage", "billing SDK" |
| Todo Tracking | "todo tracking SDK", "todo lists SDK", "task management" |
| Topic | Keywords |
|---|---|
| Structured Outputs | "structured outputs SDK", "JSON results", "validated outputs" |
| Schema Definition | "output schema SDK", "result validation" |
| Topic | Keywords |
|---|---|
| Hosting Overview | "hosting Agent SDK", "production deployment" |
| Ephemeral Sessions | "ephemeral sessions", "pattern 1 hosting" |
| Persistent Sessions | "persistent sessions", "pattern 2 hosting" |
| Topic | Keywords |
|---|---|
| Plugins | "plugins SDK", "loading plugins", "programmatic plugins" |
| MCP Integration | "MCP SDK", "MCP servers SDK", "extending with MCP" |
| Topic | Keywords |
|---|---|
| Branding Guidelines | "agent branding", "Claude Agent naming", "SDK branding" |
| Topic | Keywords |
|---|---|
| Subagents | "subagents SDK", "SDK subagents", "subagent integration" |
| Skills | "skills SDK", "SDK skills", "agent skills SDK" |
| Slash Commands | "slash commands SDK", "SDK commands", "custom commands SDK" |
| Topic | Keywords |
|---|---|
| TypeScript Reference | "TypeScript SDK reference", "TypeScript API" |
| Python Reference | "Python SDK reference", "Python API" |
| Changelog | "SDK changelog", "SDK updates" |
What do you want to do?
User asks: "How do I create a custom tool in the Agent SDK?"
1. Invoke docs-management skill
2. Use keywords: "custom tools SDK", "creating custom tools"
3. Load official documentation
4. Provide guidance based EXCLUSIVELY on official docs
User asks: "I want to build an agent with sessions, custom tools, and cost tracking"
1. Invoke docs-management skill with multiple queries:
- "session management SDK", "session resumption"
- "custom tools SDK", "creating custom tools"
- "cost tracking SDK", "token usage"
2. Synthesize guidance from official documentation
User reports: "My session resumption isn't working"
1. Invoke docs-management skill
2. Use keywords: "session management SDK", "resume option", "session ID"
3. Check official docs for session patterns
4. Guide user through proper session handling
| Issue | Keywords for docs-management |
|---|---|
| Authentication failing | "Agent SDK authentication", "ANTHROPIC_API_KEY" |
| Session not resuming | "session resumption", "resume option" |
| Custom tool not working | "custom tools SDK", "tool handlers" |
| Permissions blocking | "allowedTools", "permissionMode SDK" |
| System prompt not loading | "CLAUDE.md SDK", "settingSources" |
| Cost tracking issues | "cost tracking SDK", "token usage" |
| Structured output invalid | "structured outputs SDK", "schema definition" |
| Streaming not working | "streaming input mode", "streaming vs single" |
Skills that work well with agent-sdk-development:
Delegation example:
When working with Agent SDK + MCP integration:
1. Use agent-sdk-development for SDK-specific guidance
2. Use mcp-integration for MCP server configuration
3. Both skills delegate to docs-management for official documentation
For test scenarios, multi-model testing notes, repository-specific guidance, and external references, see:
v1.1.0 (2025-11-27): Comprehensive audit and enhancement
v1.0.0 (2025-11-26): Initial release
Date: 2025-11-28 Model: claude-opus-4-5-20251101
This skill should be used when the user asks to "create a slash command", "add a command", "write a custom command", "define command arguments", "use command frontmatter", "organize commands", "create command with file references", "interactive command", "use AskUserQuestion in command", or needs guidance on slash command structure, YAML frontmatter fields, dynamic arguments, bash execution in commands, user interaction patterns, or command development best practices for Claude Code.
This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.
This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "implement prompt-based hooks", "use ${CLAUDE_PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, SessionEnd, UserPromptSubmit, PreCompact, Notification). Provides comprehensive guidance for creating and implementing Claude Code plugin hooks with focus on advanced prompt-based hooks API.