diff --git a/docs/claude-code/auto-mode.md b/docs/claude-code/auto-mode.md index 151da87..c96e4f8 100644 --- a/docs/claude-code/auto-mode.md +++ b/docs/claude-code/auto-mode.md @@ -37,7 +37,7 @@ As of v2.1.142, Claude Code ignores `defaultMode: "auto"` in project and local s **Team/Enterprise:** An admin must enable auto mode in [Claude Code admin settings](https://claude.ai/admin-settings/claude-code) before individual users can access it. -**Bedrock / Vertex AI / Foundry:** Auto mode is off until you set the `CLAUDE_CODE_ENABLE_AUTO_MODE` environment variable to `1` (Claude Code v2.1.158 or later). See [Availability](#availability). +**Bedrock / Google Cloud's Agent Platform / Foundry:** From v2.1.207, auto mode is available by default on these providers, with no opt-in required. (In v2.1.158 through v2.1.206 it was off until you set `CLAUDE_CODE_ENABLE_AUTO_MODE=1`; that variable is still accepted for compatibility but has no effect from v2.1.207 onward.) See [Availability](#availability). ## How it works @@ -140,8 +140,8 @@ claude auto-mode critique | Requirement | Detail | |-------------|--------| | **Plans** | All plans (Pro, Max, Team, Enterprise, and the Anthropic API). On Team and Enterprise, an admin must enable it first. | -| **Models** | On the Anthropic API: Claude Opus 4.6 or later, or Sonnet 4.6. On Bedrock, Vertex AI, and Foundry: only Opus 4.7 and Opus 4.8. | -| **Providers** | Available by default on the Anthropic API. On Amazon Bedrock, Google Vertex AI, and Microsoft Foundry, set `CLAUDE_CODE_ENABLE_AUTO_MODE=1` (v2.1.158+) to enable it. | +| **Models** | On the Anthropic API: Claude Opus 4.6 or later, or Sonnet 4.6. On Bedrock, Google Cloud's Agent Platform, and Foundry: only Opus 4.7 and Opus 4.8. | +| **Providers** | Available by default on the Anthropic API, Claude Platform on AWS, Amazon Bedrock, Google Cloud's Agent Platform, and Microsoft Foundry. From v2.1.207 no opt-in is required; in v2.1.158 through v2.1.206, Bedrock, Agent Platform, and Foundry needed `CLAUDE_CODE_ENABLE_AUTO_MODE=1`. The variable is still accepted for compatibility but has no effect from v2.1.207 onward. | | **Not available** | Sonnet 4.5, Opus 4.5, Haiku, and Claude 3 models on any provider | | **Enterprise opt-out** | Admins can disable with `permissions.disableAutoMode` set to `"disable"` in managed settings | diff --git a/docs/claude-code/commands.md b/docs/claude-code/commands.md index 0007ea3..7c1ee01 100644 --- a/docs/claude-code/commands.md +++ b/docs/claude-code/commands.md @@ -21,7 +21,7 @@ Some commands are **bundled skills** rather than hard-coded CLI behavior. They u | `/exit` | Exit Claude Code (alias: `/quit`) | | `/clear` | Clear conversation history and free up context (aliases: `/reset`, `/new`) | | `/resume [session]` | Resume a previous conversation by ID or name, or open the session picker (alias: `/continue`) | -| `/fork [name]` | Create a fork of the current conversation at this point | +| `/fork [name]` | Copy the current conversation into a new background session (with its own row in `claude agents`) so it can branch off while you keep working in the original. The in-session forked subagent this used to launch is now `/subtask` | | `/rename [name]` | Rename the current session (auto-generates a name if none provided) | | `/export [filename]` | Export current conversation as plain text | | `/copy` | Copy the last assistant response to clipboard (shows interactive picker for code blocks) | @@ -86,6 +86,7 @@ Some commands are **bundled skills** rather than hard-coded CLI behavior. They u | `/agents` | Manage agent and subagent configurations | | `/tasks` | List and manage background tasks | | `/background [prompt]` | Detach the current session to run as a background agent and free the terminal. Pass a prompt to send one more instruction before detaching. Alias: `/bg` | +| `/subtask [prompt]` | Launch an in-session forked subagent that continues from the current conversation state. This is the behavior `/fork` used to have before `/fork` became a background-session copy | | `/stop` | Stop the current background session (only available while attached to one) | | `/batch ` | **[Skill]** Orchestrate large-scale changes across a codebase in parallel: decomposes the work into independent units and spawns one background subagent per unit in an isolated git worktree | | `/goal [condition]` | Set a completion goal; Claude keeps working across turns until the condition is met. `clear` or `cancel` removes the goal early | @@ -111,12 +112,13 @@ Some commands are **bundled skills** rather than hard-coded CLI behavior. They u |---------|-------------| | `/login` | Sign in to your Anthropic account | | `/logout` | Sign out from your Anthropic account | -| `/doctor` | Diagnose and verify your Claude Code installation and settings | +| `/doctor` | Run a full setup checkup (v2.1.205+): diagnoses installation health, finds unused skills, MCP servers, and plugins versus their context cost, deduplicates local `CLAUDE.md` files against checked-in ones, proposes trimming `CLAUDE.md` content Claude could derive from the codebase, and flags slow hooks. Reports findings first and asks for confirmation before changing anything. Alias: `/checkup` | | `/status` | Show version, model, account, and connectivity status | | `/stats` | Visualize daily usage, session history, streaks, and model preferences | | `/insights` | Generate a report analyzing your Claude Code sessions | | `/team-onboarding` | Generate a team onboarding guide from your Claude Code usage history | | `/powerup` | Launch interactive lessons with animated demos of Claude Code features | +| `/radio` | Open the Claude FM lo-fi radio stream in your browser (prints the stream URL when no browser is available). Not available on Amazon Bedrock, Google Cloud's Agent Platform, or Microsoft Foundry | | `/voice [hold\|tap\|off]` | Toggle voice dictation (requires a Claude.ai account) | | `/feedback [report]` | Submit feedback about Claude Code (alias: `/bug`) | | `/release-notes` | View the full changelog | diff --git a/docs/claude-code/flags.md b/docs/claude-code/flags.md index 98ca172..d9f98d6 100644 --- a/docs/claude-code/flags.md +++ b/docs/claude-code/flags.md @@ -33,7 +33,7 @@ keywords: [Claude Code CLI flags, command line options, --continue, --resume, -- | Flag | Purpose | |------|---------| -| `--permission-mode ` | Permission mode: `default`, `acceptEdits`, `plan`, `auto`, `dontAsk`, `bypassPermissions` | +| `--permission-mode ` | Permission mode: `default` (also accepted as `manual`), `acceptEdits`, `plan`, `auto`, `dontAsk`, `bypassPermissions` | | `--allowedTools ` | Allow specific tools (e.g. `"Bash(git:*) Edit Read"`) | | `--disallowedTools ` | Deny specific tools (e.g. `"Bash(rm:*)"`) | | `--tools ` | Restrict the available tool set entirely (`""` to disable all, `"default"` for all) | @@ -51,6 +51,7 @@ keywords: [Claude Code CLI flags, command line options, --continue, --resume, -- | `--print "prompt"` | Run a single prompt in headless mode (no interactive session) | | `--output-format ` | Output format with `--print`: `text`, `json`, or `stream-json` | | `--input-format ` | Input format with `--print`: `text` or `stream-json` | +| `--forward-subagent-text` | Include subagent text and thinking blocks in `stream-json` output (also settable via `CLAUDE_CODE_FORWARD_SUBAGENT_TEXT`) | | `--json-schema ` | Enforce structured output with a JSON Schema | | `--no-session-persistence` | Do not save the session to disk (headless mode only) | | `--bare` | Minimal mode: skip auto-discovery of hooks, skills, plugins, MCP servers, auto memory, and CLAUDE.md. Faster starts for scripted calls. | @@ -106,6 +107,12 @@ keywords: [Claude Code CLI flags, command line options, --continue, --resume, -- | `--verbose` | Enable verbose output | | `--safe-mode` | Launch with all customizations disabled (`CLAUDE.md`, skills, plugins, hooks, MCP servers, custom commands and agents do not load) to isolate a broken configuration. Authentication, model selection, built-in tools, and permissions still work. Also settable via the `CLAUDE_CODE_SAFE_MODE` environment variable. | +## Accessibility + +| Flag | Purpose | +|------|---------| +| `--ax-screen-reader` | Start the session in screen reader mode: replace the visual terminal interface (boxes, spinners, in-place redraws) with plain, linear labeled lines that a screen reader such as VoiceOver or NVDA reads in order. Also settable per shell with the `CLAUDE_AX_SCREEN_READER` environment variable or everywhere with the `axScreenReader` setting. | + ## Environment variables | Variable | Purpose | @@ -114,4 +121,7 @@ keywords: [Claude Code CLI flags, command line options, --continue, --resume, -- | `MAX_THINKING_TOKENS` | Control how many tokens the model uses for reasoning (set to `0` to disable thinking) | | `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` | Set to `1` to revert to a fixed thinking token budget instead of adaptive reasoning (applies to Sonnet 4.6 and Opus 4.6 only) | | `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Set to `1` to disable auto memory globally | +| `CLAUDE_CODE_MAX_WEB_SEARCHES_PER_SESSION` | Tune the per-session cap on WebSearch calls (default `200`) that stops runaway loops | +| `CLAUDE_CODE_MAX_SUBAGENTS_PER_SESSION` | Tune the per-session cap on subagent spawns (default `200`) that stops runaway loops | +| `CLAUDE_CODE_MCP_AUTO_BACKGROUND_MS` | Tune or disable the threshold after which a long-running MCP tool call moves to the background automatically (default two minutes) so the session stays usable | | `ANTHROPIC_MODEL` | Override the default model for all sessions | diff --git a/docs/claude-code/permissions.md b/docs/claude-code/permissions.md index f9ac7e4..5a7d201 100644 --- a/docs/claude-code/permissions.md +++ b/docs/claude-code/permissions.md @@ -15,13 +15,15 @@ Claude Code supports several permission modes that change how much Claude prompt | Mode | What runs without asking | Best for | |------|-------------------------|----------| -| `default` | Read-only operations | Getting started, sensitive work | +| `default` (now called **Manual**) | Read-only operations | Getting started, sensitive work | | `acceptEdits` | Reads, file edits, and common filesystem commands (`mkdir`, `touch`, `mv`, `cp`, etc.) | Iterating on code you're reviewing | | `plan` | Read-only (Claude analyzes but cannot modify files) | Exploring a codebase before changing it | | `auto` | Everything, with background safety checks | Long tasks, reducing prompt fatigue | | `dontAsk` | Only pre-approved tools | Locked-down CI and scripts | | `bypassPermissions` | Everything, including protected paths | Isolated containers and VMs only | +The `default` mode is now labelled **Manual** across the CLI, `--help`, VS Code, and JetBrains, and `--permission-mode manual` is accepted as an alias for `default`. From v2.1.203, the status bar shows a gray `⏸ manual mode on` badge while it is active. The stored setting value remains `default`. + Set a default mode in your settings: ```json diff --git a/docs/claude-code/subagents.md b/docs/claude-code/subagents.md index ed0360c..15031b3 100644 --- a/docs/claude-code/subagents.md +++ b/docs/claude-code/subagents.md @@ -65,17 +65,13 @@ Output a structured report with severity ratings (critical / high / medium / low | `model` | No | Model to use: `sonnet`, `opus`, `haiku`, a full model ID, or `inherit`. Defaults to `inherit` (uses the parent session's model) | | `tools` | No | Comma-separated list of allowed tools; omit to inherit all tools from the parent session | | `disallowedTools` | No | Tools to deny, removed from the inherited or specified list | -| `permissionMode` | No | Permission mode for this subagent: `default`, `acceptEdits`, `plan`, `auto`, `dontAsk`, or `bypassPermissions` | +| `permissionMode` | No | Permission mode for this subagent: `default` (also accepted as `manual`), `acceptEdits`, `plan`, `auto`, `dontAsk`, or `bypassPermissions` | | `maxTurns` | No | Maximum number of agentic turns before the subagent stops | | `skills` | No | Skills to load into the subagent's context at startup. Full skill content is injected; subagents do not inherit skills from the parent session | | `mcpServers` | No | MCP servers available to this subagent. Each entry is a server name referencing an already-configured server or an inline server definition | | `hooks` | No | Lifecycle hooks scoped to this subagent | | `memory` | No | Persistent memory scope: `user`, `project`, or `local`. Enables cross-session learning for this subagent | -| `background` | No | Set to `true` to always run this subagent as a background task. Default: `false` | -| `effort` | No | Effort level: `low`, `medium`, `high`, `xhigh`, or `max`. Overrides the session effort level for this subagent | -| `isolation` | No | Set to `worktree` to run the subagent in a temporary git worktree, giving it an isolated copy of the repository | -| `color` | No | Display color in the task list: `red`, `blue`, `green`, `yellow`, `purple`, `orange`, `pink`, or `cyan` | -| `initialPrompt` | No | Auto-submitted as the first user turn when this agent runs as the main session agent (via `--agent` CLI flag) | +| `background` | No | Set to `true` to always run this subagent as a background task, even when the main agent needs its result right away. When unset, Claude chooses, and as of v2.1.198 it runs subagents in the background by default | | `effort` | No | Effort level: `low`, `medium`, `high`, `xhigh`, or `max`. Overrides the session effort level for this subagent | | `isolation` | No | Set to `worktree` to run the subagent in a temporary git worktree, giving it an isolated copy of the repository | | `color` | No | Display color in the task list: `red`, `blue`, `green`, `yellow`, `purple`, `orange`, `pink`, or `cyan` | @@ -722,13 +718,13 @@ Run the test-runner agent on the files I just edited, then continue. ### Background vs. foreground -By default, subagents run in the foreground; the main agent waits for a result before continuing. For independent tasks you can run them in the background: +As of v2.1.198, subagents run in the **background by default**: the main agent keeps working and picks up the result when the subagent finishes. The main agent still runs a subagent in the foreground when it needs the result before it can continue. You can also steer this explicitly: ``` Run the test suite in the background while I keep working. ``` -The main agent will notify you when a background subagent completes. +Press **Ctrl+B** to background a running task. A background subagent's result reaches the main agent as a completion notification in a later turn (v2.1.211+); if you ask about progress before it finishes, the main agent reports that the subagent is still running rather than guessing at results. When a background subagent needs a permission decision, the prompt surfaces in your main session instead of being auto-denied. The dialog shows which agent is asking, and pressing **Esc** denies only that one tool call.