# Atomic

## Docs

- [How the agent loop works](https://docs.bastani.ai/agents/claude-code/agent-sdk/agent-loop.md): Understand the message lifecycle, tool execution, context window, and architecture that power your SDK agents.
- [Use Claude Code features in the SDK](https://docs.bastani.ai/agents/claude-code/agent-sdk/core-concepts/claude-code-features.md): Load project instructions, skills, hooks, and other Claude Code features into your SDK agents.
- [Work with sessions](https://docs.bastani.ai/agents/claude-code/agent-sdk/core-concepts/sessions.md): How sessions persist agent conversation history, and when to use continue, resume, and fork to return to a prior run.
- [Track cost and usage](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/cost-tracking.md): Track token usage, deduplicate parallel tool calls, and calculate costs with the Claude Agent SDK.
- [Give Claude custom tools](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/custom-tools.md): Define custom tools with the Agent SDK's in-process MCP server so Claude can call your functions and APIs.
- [Rewind file changes with checkpointing](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/file-checkpointing.md): Track file changes during agent sessions and restore files to any previous state.
- [Intercept and control agent behavior with hooks](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/hooks.md): Intercept and customize agent behavior at key execution points with hooks in the Claude Agent SDK.
- [Hosting the Agent SDK](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/hosting.md): Deploy and host the Claude Agent SDK in production environments.
- [Connect to external tools with MCP](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/mcp.md): Configure MCP servers to extend your agent with external tools, covering transports, tool search, authentication, and error handling.
- [Modifying system prompts](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/modifying-system-prompts.md): Customize Claude's behavior by modifying system prompts using output styles, append mode, or a custom system prompt.
- [Configure permissions](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/permissions.md): Control how your agent uses tools with permission modes, hooks, and declarative allow/deny rules.
- [Plugins in the SDK](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/plugins.md): Load custom plugins to extend Claude Code with commands, agents, skills, and hooks through the Agent SDK.
- [Securely deploying AI agents](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/secure-deployment.md): Secure Claude Code and Agent SDK deployments with isolation, credential management, and network controls.
- [Agent Skills in the SDK](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/skills.md): Extend Claude with specialized capabilities using Agent Skills in the Claude Agent SDK.
- [Slash commands in the SDK](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/slash-commands.md): Use slash commands to control Claude Code sessions through the SDK.
- [Stream responses in real-time](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/streaming-output.md): Get real-time responses from the Agent SDK as text and tool calls stream in.
- [Streaming input](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/streaming-vs-single-mode.md): Understand the two input modes for the Claude Agent SDK and when to use streaming versus single-message mode.
- [Structured outputs](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/structured-output.md): Get type-safe, schema-validated JSON results from the Claude Agent SDK.
- [Subagents in the SDK](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/subagents.md): Define and invoke subagents to isolate context, run tasks in parallel, and apply specialized instructions.
- [Todo lists](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/todo-tracking.md): Track and display todos using the Claude Agent SDK for organized task management.
- [Scale to many tools with tool search](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/tool-search.md): Scale your agent to thousands of tools by discovering and loading only what's needed, on demand.
- [Handle approvals and user input](https://docs.bastani.ai/agents/claude-code/agent-sdk/guides/user-input.md): Surface Claude's approval requests and clarifying questions to users, then return their decisions to the SDK.
- [Headless mode](https://docs.bastani.ai/agents/claude-code/agent-sdk/headless.md): Run Claude Code non-interactively from scripts and automation using headless mode and the print flag.
- [Agent SDK overview](https://docs.bastani.ai/agents/claude-code/agent-sdk/overview.md): Build production AI agents with Claude Code as a library, using the same tools, agent loop, and context management.
- [Quickstart](https://docs.bastani.ai/agents/claude-code/agent-sdk/quickstart.md): Get started with the Python or TypeScript Agent SDK to build AI agents that work autonomously.
- [TypeScript SDK reference](https://docs.bastani.ai/agents/claude-code/agent-sdk/sdk-references/typescript.md): Complete API reference for the Claude Agent SDK TypeScript package, including functions, types, and options.
- [CLI reference](https://docs.bastani.ai/agents/claude-code/cli/cli-reference.md): Complete reference for the Claude Code command-line interface, including commands, flags, and configuration options.
- [Hooks reference](https://docs.bastani.ai/agents/claude-code/cli/hooks.md): Configure shell commands that execute at points in the Claude Code lifecycle to customize and extend agent behavior.
- [Permissions](https://docs.bastani.ai/agents/claude-code/cli/permissions.md): How Claude Code's permission system controls tool access, including allow/deny/ask rules and permission modes.
- [Agent Skills](https://docs.bastani.ai/agents/claude-code/cli/skills.md): Package expertise into discoverable Skills that Claude Code loads on demand to perform specialized tasks.
- [Subagents](https://docs.bastani.ai/agents/claude-code/cli/subagents.md): Create and configure specialized subagents that handle dedicated tasks with isolated context in Claude Code.
- [Tools available to Claude](https://docs.bastani.ai/agents/claude-code/cli/tools.md): Reference for the built-in tools Claude Code can use, including file operations, command execution, and web access.
- [Hooks configuration](https://docs.bastani.ai/agents/copilot-cli/hooks.md): Configure hooks for use with the GitHub Copilot CLI and Copilot coding agent.
- [Copilot SDK for Node.js and TypeScript](https://docs.bastani.ai/agents/copilot-cli/sdk.md): TypeScript SDK for programmatic control of the GitHub Copilot CLI via JSON-RPC.
- [About Agent Skills](https://docs.bastani.ai/agents/copilot-cli/skills.md): How Agent Skills enhance the GitHub Copilot CLI and coding agent with specialized capabilities.
- [Custom agents configuration](https://docs.bastani.ai/agents/copilot-cli/subagents.md): Reference for configuring custom agents in the GitHub Copilot CLI.
- [GitHub Copilot CLI command reference](https://docs.bastani.ai/agents/copilot-cli/usage.md): Commands and keyboard shortcuts for using the GitHub Copilot CLI effectively.
- [SDK](https://docs.bastani.ai/agents/opencode/sdk.md): Type-safe JavaScript client for the opencode server.
- [Server](https://docs.bastani.ai/agents/opencode/server.md): Interact with the opencode server over HTTP using the headless serve command and OpenAPI endpoint.
- [CLI commands](https://docs.bastani.ai/cli/commands.md): Every atomic subcommand — chat, workflow, session, completions, update, uninstall.
- [Context files](https://docs.bastani.ai/cli/context-files.md): Generate CLAUDE.md and AGENTS.md so every future session starts with the right context.
- [Agents](https://docs.bastani.ai/concepts/agents.md): Multi-agent support — Claude Code, OpenCode, and GitHub Copilot CLI behind one workflow runtime.
- [CLI vs SDK](https://docs.bastani.ai/concepts/cli-vs-sdk.md): Two surfaces over one workflow runtime — and how to choose between them.
- [Permissions](https://docs.bastani.ai/concepts/permissions.md): Atomic disables agent permission prompts during workflows. Run in a devcontainer or worktree.
- [Sessions](https://docs.bastani.ai/concepts/sessions.md): Isolated, detachable tmux sessions on a dedicated socket — surviving disconnects.
- [Skills](https://docs.bastani.ai/concepts/skills.md): Structured capability modules that give agents best practices and reusable workflows.
- [Specialized sub-agents](https://docs.bastani.ai/concepts/sub-agents.md): Purpose-built agents with scoped context, tools, and termination conditions.
- [Why Atomic](https://docs.bastani.ai/concepts/why-atomic.md): What problem workflows solve, and how Atomic turns engineering process into code.
- [Workflows](https://docs.bastani.ai/concepts/workflows.md): Stages, the execution graph, session isolation, transcript hand-offs, and human gates.
- [Agent files](https://docs.bastani.ai/configuration/agent-files.md): How Atomic reads and writes .claude, .opencode, and .github configuration.
- [CI/CD pipeline](https://docs.bastani.ai/configuration/ci.md): The GitHub Actions workflows that build, validate, and publish Atomic.
- [Settings](https://docs.bastani.ai/configuration/settings.md): Configure the Atomic CLI via .atomic/settings.json.
- [FAQ](https://docs.bastani.ai/faq.md): Common questions and how Atomic compares to adjacent tools.
- [Introduction](https://docs.bastani.ai/index.md): Turn coding agents into reliable engineering workflows.
- [Installation](https://docs.bastani.ai/installation.md): Install the Atomic CLI, the SDK, or both.
- [Quickstart](https://docs.bastani.ai/quickstart.md): Generate context files, run Ralph, and build your first workflow.
- [defineWorkflow](https://docs.bastani.ai/sdk/define-workflow.md): The builder chain — name, agent, run, compile.
- [Embedding](https://docs.bastani.ai/sdk/embedding.md): Build your own atomic-powered app — registries, multi-workflow CLIs, and parent CLI embedding.
- [Runnable examples](https://docs.bastani.ai/sdk/examples.md): Small, complete user apps that demonstrate each SDK primitive.
- [hostLocalWorkflows](https://docs.bastani.ai/sdk/host-local-workflows.md): The explicit handoff that makes your CLI atomic-dispatchable.
- [Inputs](https://docs.bastani.ai/sdk/inputs.md): Declared input schemas, CLI flag validation, and input precedence.
- [SDK overview](https://docs.bastani.ai/sdk/overview.md): @bastani/atomic-sdk/workflows — pure primitives for authoring, registering, and running workflows.
- [WorkflowPicker](https://docs.bastani.ai/sdk/picker.md): The interactive picker as a reusable component.
- [runWorkflow](https://docs.bastani.ai/sdk/run-workflow.md): Spawn the orchestrator session, attach or detach, and embed under any parent CLI.
- [Stages](https://docs.bastani.ai/sdk/stages.md): ctx.stage, parallel fan-out, transcript hand-offs, return values, and headless stages.
- [Troubleshooting](https://docs.bastani.ai/troubleshooting.md): Fixes for common Atomic setup issues.
- [Building a workflow](https://docs.bastani.ai/workflows/building.md): Author your first workflow with the SDK — define, stage, run.
- [Built-in workflows](https://docs.bastani.ai/workflows/built-in.md): Ralph, deep-research-codebase, and open-claude-design — the workflows Atomic ships.
- [Registering workflows](https://docs.bastani.ai/workflows/registering.md): Make your own workflows discoverable through atomic workflow.
- [Running workflows](https://docs.bastani.ai/workflows/running.md): List, pick, pass inputs, detach, attach, and read run artifacts.

## OpenAPI Specs

- [openapi](https://docs.bastani.ai/api-reference/openapi.json)

## Optional

- [GitHub](https://github.com/flora131/atomic)
- [DeepWiki](https://deepwiki.com/flora131/atomic)