Claude product ecosystem — Library index
Canonical reference for everything Anthropic ships under the Claude brand: the underlying models, the Messages API + SDKs, Claude Code (CLI + IDE + web + desktop), the Claude Agent SDK, the Model Context Protocol (MCP), and the auxiliary APIs (Batches, Files, Citations, Vision, Tool Use, Memory, Web Search, Code Execution, Computer Use). House style and source URLs match the rest of the Compute library; cite specific notes from any agentic coding session that touches Claude as the runtime.
This sub-library lives inside Math-and-Compute/Compute/ as a sibling of Tier3/ and the operating-systems / databases / ML notes. It is intentionally fact-dense and version-pinned: every non-obvious claim cites either docs.claude.com, platform.claude.com, code.claude.com, modelcontextprotocol.io, or an anthropics/* GitHub repo. Where the official docs are silent but the behavior is observable (CLI flags surfaced by claude --help, undocumented env vars referenced in source, etc.), the entry is labelled “Undocumented as of 2026-05-25 — observed behavior” so a future reader knows to verify.
The contents are organized as a single flat folder of notes — no deeper nesting — because most lookups are “what does the X API parameter do” or “what does this CLI flag do”, and a flat folder collapses to a single Obsidian search.
See also
- prompt-engineering-agent-systems — vendor-neutral discipline overview (this library is the Claude-specific deep dive)
- transformer-architecture — the underlying transformer substrate Claude is built on
- fine-tuning-rlhf — how the base model gets post-trained into Claude
- inference-optimization — what the serving stack does under the hood (prompt caching, KV-cache reuse, batching)
- rag-embeddings-vector-search — pairs with
citations-and-grounding.mdfor RAG patterns - observability-stack — Claude Code emits OpenTelemetry; relevant for production deployments
Library contents
Foundations
| File | What it covers |
|---|---|
| claude-models-and-capabilities | Opus 4.7 / Sonnet 4.6 / Haiku 4.5 + legacy generations, pricing, context windows, knowledge cutoffs, benchmarks, model-selection heuristics |
| claude-api-and-sdks | The Messages API surface (parameters, response shapes, streaming, errors, rate limits) + Python / TypeScript / Go / Java / C# / PHP / Ruby SDKs |
| prompt-engineering-for-claude | Claude-specific prompting (XML tags, prefill, role priming, model-version idiosyncrasies, agentic prompting) |
| prompt-caching | 5-minute and 1-hour ephemeral caches, breakpoint placement, cost math, multi-turn caching strategy |
Tool use & multimodal
| File | What it covers |
|---|---|
| tool-use-and-function-calling | Client-defined tools (tool_use / tool_result), parallel tool use, strict mode, fine-grained tool streaming, server tools overview |
| vision-and-multimodal | Image input (base64, URL, Files API), PDF input, document blocks, token costs, sizing rules, format support |
| computer-use-and-code-execution | The four Anthropic-managed tools that make Claude agentic: computer_*, code_execution_*, bash_20250124, text_editor_*, plus memory_20250818 and web_search_* and web_fetch_* |
| batch-api-and-files-api | Message Batches (asynchronous, ≤24 h, 50 % discount) + Files API (upload, download, reference by file_id) |
| citations-and-grounding | Native citation blocks, document chunking, char_location / page_location / content_block_location, RAG integration |
Claude Code
| File | What it covers |
|---|---|
| claude-code-cli | The claude binary — install paths, flags, subcommands, settings.json, environment variables, permission modes, statusline, login |
| claude-code-hooks-deep | Every hook event, matcher syntax, command vs HTTP vs MCP-tool vs prompt vs agent handlers, exit-code semantics, JSON output schema, safety floor |
| claude-code-subagents-and-skills | Subagents (.claude/agents/*.md), the Task / Agent tool, Skills (SKILL.md), bundled vs user-created skills, hierarchy, disable-model-invocation, dynamic !backtick` |
Beyond the CLI
| File | What it covers |
|---|---|
| mcp-protocol | The Model Context Protocol spec: JSON-RPC base, lifecycle, capabilities, tools / resources / prompts primitives, transports (stdio, HTTP, SSE, streamable HTTP), auth, SDK survey |
| claude-agent-sdk | claude-agent-sdk (Python + TS) — query, ClaudeSDKClient, ClaudeAgentOptions, setting_sources, mcp_servers, agents map, hooks, message types |
| agents-and-orchestration-patterns | Multi-agent orchestration (supervisor-worker, swarm, ReAct, Plan-and-Solve), handoff protocols, Anthropic’s own published guidance |
| hidden-tricks-and-gotchas | Undocumented flags, environment-variable escape hatches, prompt-caching breakpoint tricks, tool-use parallelism quirks, vision token accounting, MCP buffering, model-specific quirks |
How to use this library
- Picking a model for a new project — start at claude-models-and-capabilities for the matrix, then prompt-caching for the cost model, then tool-use-and-function-calling if you need tools.
- Debugging a Claude Code session — claude-code-cli for flags, claude-code-hooks-deep for hook lifecycle, hidden-tricks-and-gotchas for env-var overrides.
- Building an MCP server — mcp-protocol first, then claude-code-cli for how Claude Code consumes it, then claude-agent-sdk if you want to embed an agent.
- Designing a multi-agent system — agents-and-orchestration-patterns is the playbook; claude-code-subagents-and-skills for the in-process building blocks, claude-agent-sdk for cross-process.
- Cost optimization — prompt-caching + batch-api-and-files-api + the pricing tables in claude-models-and-capabilities.
- RAG / grounded responses — citations-and-grounding + vision-and-multimodal for PDF document blocks.
Conventions used in this library
- Model IDs are quoted verbatim (e.g.
claude-opus-4-7,claude-haiku-4-5-20251001). Anthropic switched to dateless pinned IDs starting with the 4.6 generation per models overview; for 4.5-and-earlier the alias is a convenience pointer that resolves to a dated ID. - Beta header names are quoted verbatim from the
anthropic-betaenum (e.g.extended-cache-ttl-2025-04-11,computer-use-2025-11-24,skills-2025-10-02). The authoritative list is in the Files API beta header enum. - Pricing is in USD per million tokens (MTok). Where multiple
\$symbols appear on the same line they are backslash-escaped to avoid Obsidian MathJax interpretation. - Code examples are minimal and runnable, in Python + TypeScript where the SDKs differ in shape.
- “Per docs.claude.com 2026-05” tags a claim sourced from official documentation as of the file’s
createddate; “per anthropic/X v…” tags a claim sourced from a specific repo release.
Sources read while building this library (2026-05-25 pass)
platform.claude.com/docs/en/about-claude/models/overview(full model matrix + legacy table)platform.claude.com/docs/en/api/messages(Messages API ref)platform.claude.com/docs/en/build-with-claude/prompt-caching(cache mechanics)platform.claude.com/docs/en/build-with-claude/tool-use/overview(tool use base)platform.claude.com/docs/en/build-with-claude/vision(vision + image limits)platform.claude.com/docs/en/build-with-claude/extended-thinking(extended + adaptive thinking)platform.claude.com/docs/en/build-with-claude/batch-processing(batches API)platform.claude.com/docs/en/build-with-claude/citations(citations + document blocks)platform.claude.com/docs/en/api/files-content(Files API + full beta header enum)platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-toolplatform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-toolplatform.claude.com/docs/en/agents-and-tools/tool-use/memory-toolplatform.claude.com/docs/en/agents-and-tools/tool-use/web-search-toolplatform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practicescode.claude.com/docs/en/overview(Claude Code root)code.claude.com/docs/en/cli-referencecode.claude.com/docs/en/settingscode.claude.com/docs/en/slash-commandscode.claude.com/docs/en/hookscode.claude.com/docs/en/subagentscode.claude.com/docs/en/skillscode.claude.com/docs/en/mcpcode.claude.com/docs/en/agent-sdk/overviewmodelcontextprotocol.io/specification/2025-06-18/basicmodelcontextprotocol.io/specification/2025-06-18/server/toolsmodelcontextprotocol.io/specification/2025-06-18/server/resourcesmodelcontextprotocol.io/specification/2025-06-18/server/promptsgithub.com/anthropics/claude-agent-sdk-python+…-typescriptgithub.com/anthropics/anthropic-sdk-python+…-typescriptgithub.com/anthropics/claude-code(releases, source structure)
Sections that required inference (rather than direct doc lookup) are flagged in each file under “Undocumented as of 2026-05-25 — observed behavior” so they can be re-verified later.