Claude docs changes for April 9th, 2026 [diff]

Executive Summary

• New claude-mythos-preview model added to the API: "New class of intelligence, strongest in coding and cybersecurity"
• New Managed Agents API (beta): full CRUD API for Agents, Sessions, Environments, and Vaults — enabling stateful, long-running agent sessions in cloud containers
• New ant CLI added as a first-class SDK: terminal client for the Claude API with typed flags, response transforms, pagination, and native Claude Code integration
• Managed Agents pricing documented: $0.08/session-hour for runtime, plus standard token pricing (replaces Code Execution container-hour billing)
• Claude Code 2.1.97 brings many fixes and improvements including a focus view toggle, status line refresh interval, live subagent indicators, and dozens of bug fixes

New Claude Code versions

2.1.97

New features

• Added focus view toggle (Ctrl+O) in NO_FLICKER mode showing prompt, one-line tool summary with edit diffstats, and final response
• Added refreshInterval status line setting to re-run the status line command every N seconds
• Added workspace.git_worktree to the status line JSON input, set when the current directory is inside a linked git worktree
• Added ● N running indicator in /agents next to agent types with live subagent instances
• Added syntax highlighting for Cedar policy files (.cedar, .cedarpolicy)

Existing feature improvements

• Improved Accept Edits mode to auto-approve filesystem commands prefixed with safe env vars or process wrappers (e.g. LANG=C rm foo, timeout 5 mkdir out)
• Improved auto mode and bypass-permissions mode to auto-approve sandbox network access prompts
• Improved sandbox: sandbox.network.allowMachLookup now takes effect on macOS
• Improved image handling: pasted and attached images are now compressed to the same token budget as images read via the Read tool
• Improved slash command and @-mention completion to trigger after CJK sentence punctuation
• Improved Bridge sessions to show the local git repo, branch, and working directory on the claude.ai session card
• Improved footer layout: indicators (Focus, notifications) now stay on the mode-indicator row instead of wrapping below
• Improved context-low warning to show as a transient footer notification instead of a persistent row
• Improved markdown blockquotes to show a continuous left bar across wrapped lines
• Improved session transcript size by skipping empty hook entries and capping stored pre-edit file copies
• Improved transcript accuracy: per-block entries now carry the final token usage instead of the streaming placeholder
• Improved Bash tool OTEL tracing: subprocesses now inherit a W3C TRACEPARENT env var when tracing is enabled
• Updated /claude-api skill to cover Managed Agents alongside the Claude API

Major bug fixes

• Fixed --dangerously-skip-permissions being silently downgraded to accept-edits mode after approving a write to a protected path
• Fixed Bash tool permissions, tightening checks around env-var prefixes and network redirects, and reducing false prompts
• Fixed permission rules with names matching JavaScript prototype properties (e.g. toString) causing settings.json to be silently ignored
• Fixed managed-settings allow rules remaining active after an admin removed them until process restart
• Fixed permissions.additionalDirectories changes in settings not applying mid-session
• Fixed MCP HTTP/SSE connections accumulating ~50 MB/hr of unreleased buffers when servers reconnect
• Fixed MCP OAuth oauth.authServerMetadataUrl not being honored on token refresh after restart, fixing ADFS and similar IdPs
• Fixed 429 retries burning all attempts in ~13 seconds — exponential backoff now applies as a minimum
• Fixed rate-limit upgrade options disappearing after context compaction
• Fixed several /resume picker issues (uneditable sessions, search wipe on reload, empty list navigation, and more)
• Fixed file-edit diffs disappearing on --resume when the edited file was larger than 10KB
• Fixed messages typed while Claude is working not being persisted to the transcript
• Fixed Stop/SubagentStop hooks failing on long sessions
• Fixed subagents with worktree isolation leaking their working directory back to the parent session's Bash tool
• Fixed compaction writing duplicate multi-MB subagent transcript files on prompt-too-long retries
• Fixed claude plugin update incorrectly reporting "already at the latest version" for git-based marketplace plugins
• Fixed slash command picker breaking when a plugin's name is a YAML boolean keyword
• Fixed Korean/Japanese/Unicode text becoming garbled when copied in no-flicker mode on Windows
• Fixed Bedrock SigV4 authentication failing when AWS_BEARER_TOKEN_BEDROCK or ANTHROPIC_BEDROCK_BASE_URL are set to empty strings
• Fixed several NO_FLICKER mode issues: scroll artifacts, crash on MCP hover, memory leak on retries, slow mouse-wheel scrolling on Windows Terminal, missing custom status line on short terminals, Shift+Enter/Alt+Cmd+arrow shortcuts in Warp

Claude Code changes

No new or changed Claude Code documentation files.

API changes

New Documents

API/beta/agents [Source]

New Managed Agents API reference covering full CRUD operations for Agent resources: Create, List, Get, Update, Archive, and List Versions. Agents define the model, system prompt, tools (built-in toolsets, MCP toolsets, custom tools), skills, and MCP server connections used by sessions.

API/beta/environments [Source]

New Environments API reference for creating and managing cloud execution environments for agent sessions. Environments configure networking (unrestricted or limited with allowed hosts, MCP servers, and package managers) and pre-installed packages (apt, pip, npm, go, cargo, gem).

API/beta/sessions [Source]

New Sessions API reference for stateful agent sessions running in cloud containers. Covers Create, List, Get, Update, Delete, and Archive operations, plus sub-resources: Events (list, send, stream), Resources (file and GitHub repository mounts).

API/beta/vaults [Source]

New Vaults API reference for securely storing credentials used by agents during sessions. Vaults hold credentials with MCP OAuth (including refresh token support) or static bearer token authentication types. Sub-resource: Credentials (Create, List, Get, Update, Delete, Archive).

API/sdks/cli [Source]

New documentation for the ant CLI — a terminal client for the Claude API. Covers installation (Homebrew, curl, Go), authentication, command structure (resource action pattern with beta: prefix for beta resources), output formats (JSON, YAML, JSONL, interactive explorer TUI), GJSON transforms, request body input (flags, stdin, @file references), version-controlling API resources, scripting patterns, and shell completion. Claude Code understands how to use ant natively.

Changed documents

about-claude/models/migration-guide [Source]

• Added note that the guide covers the Messages API and points Managed Agents users to a separate migration guide. [line 4] [Source]

about-claude/models/whats-new-claude-4-6 [Source]

• Added language labels (Python, Shell) to code blocks for clarity.

about-claude/pricing [Source]

• Replaced "Agent use case pricing examples" section with "Claude Managed Agents pricing" section. Documents two billing dimensions: tokens (at standard rates; batch API, fast mode, data residency, long context, and third-party platform modifiers do not apply) and session runtime ($0.08/session-hour while session is in running status). Includes a worked example for a one-hour coding session. [line 282+]

agents-and-tools/agent-skills/best-practices [Source]

• Removed links to Agent SDK skills documentation from navigation footer.

agents-and-tools/agent-skills/enterprise [Source]

• Removed link to Agent SDK secure deployment documentation from navigation footer.

agents-and-tools/agent-skills/overview [Source]

• Removed "Claude Agent SDK" section describing how Skills work in the Agent SDK. Navigation footer link to Agent SDK skills also removed.

agents-and-tools/agent-skills/quickstart [Source]

• Updated code examples from Python SDK to CLI (ant) for listing skills and creating messages with skills.

agents-and-tools/mcp-connector [Source]

• Removed mention of Agent SDK from guidance on when to use client-side MCP helpers vs. the mcp_servers API parameter.

api/beta [Source]

• Added new model claude-mythos-preview to the model enum, described as "New class of intelligence, strongest in coding and cybersecurity". [line 7155+]
• Removed user-profiles-2026-03-24 beta header from the list of available beta features.

api/beta-headers [Source]

• Added "Endpoint-specific headers" section documenting the managed-agents-2026-04-01 beta header required for all Managed Agents endpoints (/v1/agents, /v1/sessions, /v1/environments). [line 53+]
• Updated the code example in the quick-start from Python to cURL.

api/client-sdks [Source]

• Added the ant CLI as the first SDK listed, with the description "Shell scripting, typed flags, response transforms". Updated quick installation and quick start examples to use CLI instead of Python.

api/errors [Source]

• Updated request-ID example from Python SDK to the ant CLI (ant --debug).