--- title: Create Agent url: https://platform.claude.com/docs/en/api/java/beta/agents/create --- # Create Agent `BetaManagedAgentsAgent beta().agents().create(params, requestOptions = RequestOptions.none())` **POST** `/v1/agents` Create Agent ## Parameters - `AgentCreateParams params` - `Optional> betas` (header parameter) This endpoint is in beta: requests must send `managed-agents-2026-04-01` in this header. - `MESSAGE_BATCHES_2024_09_24("message-batches-2024-09-24")` - `PROMPT_CACHING_2024_07_31("prompt-caching-2024-07-31")` - `COMPUTER_USE_2024_10_22("computer-use-2024-10-22")` - `COMPUTER_USE_2025_01_24("computer-use-2025-01-24")` - `PDFS_2024_09_25("pdfs-2024-09-25")` - `TOKEN_COUNTING_2024_11_01("token-counting-2024-11-01")` - `TOKEN_EFFICIENT_TOOLS_2025_02_19("token-efficient-tools-2025-02-19")` - `OUTPUT_128K_2025_02_19("output-128k-2025-02-19")` - `FILES_API_2025_04_14("files-api-2025-04-14")` - `MCP_CLIENT_2025_04_04("mcp-client-2025-04-04")` - `MCP_CLIENT_2025_11_20("mcp-client-2025-11-20")` - `DEV_FULL_THINKING_2025_05_14("dev-full-thinking-2025-05-14")` - `INTERLEAVED_THINKING_2025_05_14("interleaved-thinking-2025-05-14")` - `CODE_EXECUTION_2025_05_22("code-execution-2025-05-22")` - `EXTENDED_CACHE_TTL_2025_04_11("extended-cache-ttl-2025-04-11")` - `CONTEXT_1M_2025_08_07("context-1m-2025-08-07")` - `CONTEXT_MANAGEMENT_2025_06_27("context-management-2025-06-27")` - `MODEL_CONTEXT_WINDOW_EXCEEDED_2025_08_26("model-context-window-exceeded-2025-08-26")` - `SKILLS_2025_10_02("skills-2025-10-02")` - `FAST_MODE_2026_02_01("fast-mode-2026-02-01")` - `OUTPUT_300K_2026_03_24("output-300k-2026-03-24")` - `USER_PROFILES_2026_03_24("user-profiles-2026-03-24")` - `USER_PROFILES_2026_08_18("user-profiles-2026-08-18")` - `USER_PROFILES_2026_09_04("user-profiles-2026-09-04")` - `ADVISOR_TOOL_2026_03_01("advisor-tool-2026-03-01")` - `MANAGED_AGENTS_2026_04_01("managed-agents-2026-04-01")` - `CACHE_DIAGNOSIS_2026_04_07("cache-diagnosis-2026-04-07")` - `DREAMING_2026_04_21("dreaming-2026-04-21")` - `THINKING_TOKEN_COUNT_2026_05_13("thinking-token-count-2026-05-13")` - `SERVER_SIDE_FALLBACK_2026_06_01("server-side-fallback-2026-06-01")` - `SERVER_SIDE_FALLBACK_2026_07_01("server-side-fallback-2026-07-01")` - `FALLBACK_CREDIT_2026_06_01("fallback-credit-2026-06-01")` - `FALLBACK_CREDIT_2026_07_01("fallback-credit-2026-07-01")` - `AGENT_MEMORY_2026_07_22("agent-memory-2026-07-22")` - `MID_CONVERSATION_TOOL_CHANGES_2026_07_01("mid-conversation-tool-changes-2026-07-01")` - `COMPACT_2026_01_12("compact-2026-01-12")` - `COMPUTER_USE_2025_11_24("computer-use-2025-11-24")` - `MCP_TUNNELS_2026_06_22("mcp-tunnels-2026-06-22")` - `STRUCTURED_OUTPUTS_2025_11_13("structured-outputs-2025-11-13")` - `TASK_BUDGETS_2026_03_13("task-budgets-2026-03-13")` - `THINKING_DISPLAY_UPDATES_2026_08_18("thinking-display-updates-2026-08-18")` - `CE_USER_MANAGEMENT_2026_07_13("ce-user-management-2026-07-13")` - `MID_CONVERSATION_OUTPUT_CONFIG_2026_07_01("mid-conversation-output-config-2026-07-01")` - `THINKING_BINDING_CONTROLS_2026_08_01("thinking-binding-controls-2026-08-01")` - `MID_CONVERSATION_SYSTEM_CLEAR_AT_2026_08_21("mid-conversation-system-clear-at-2026-08-21")` - `COMPACT_2026_09_04("compact-2026-09-04")` - `INLINE_TOOLS_2026_09_15("inline-tools-2026-09-15")` - `MCP_CLIENT_2026_09_15("mcp-client-2026-09-15")` - `CE_PLUGINS_2026_09_01("ce-plugins-2026-09-01")` - `SPEND_LIMIT_READS_2026_09_26("spend-limit-reads-2026-09-26")` - `TELEMETRY_DESTINATIONS_2026_08_11("telemetry-destinations-2026-08-11")` - `Optional workspaceId` (header parameter) Optional header to select the Workspace for this request. The value is a Workspace ID (for example, `wrkspc_011CZkZaBF1tNoB5wlCeusgy`). Only needed for credentials that can act on more than one Workspace. A credential that belongs to a specific Workspace may omit it; if sent, it must match that Workspace. - `Model model` Model identifier. Accepts the [model string](https://platform.claude.com/docs/en/about-claude/models/overview?sdk_languages=java#latest-models-comparison), e.g. `claude-opus-5-5`, or a `model_config` object for additional configuration control - `enum BetaManagedAgentsModel` The model that will power your agent. See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. - `CLAUDE_HAIKU_5_5("claude-haiku-5-5")` Fastest model for high-volume, real-time tasks - `CLAUDE_SONNET_5_5("claude-sonnet-5-5")` Efficient model for coding and agents - `CLAUDE_OPUS_5_5("claude-opus-5-5")` Powerful intelligence for coding, knowledge work, and long-running agents - `CLAUDE_FABLE_5_1("claude-fable-5-1")` Frontier intelligence for ambitious tasks across coding, scientific discovery, and enterprise workflows - `CLAUDE_SONNET_5("claude-sonnet-5")` Efficient model for coding and agents - `CLAUDE_FABLE_5("claude-fable-5")` Next generation of intelligence for the hardest knowledge work and coding problems - `CLAUDE_OPUS_5("claude-opus-5")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_8("claude-opus-4-8")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_7("claude-opus-4-7")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_6("claude-opus-4-6")` Powerful intelligence for long-running agents and coding - `CLAUDE_SONNET_4_6("claude-sonnet-4-6")` Best combination of speed and intelligence - `CLAUDE_HAIKU_4_5("claude-haiku-4-5")` - `CLAUDE_HAIKU_4_5_20251001("claude-haiku-4-5-20251001")` - `CLAUDE_OPUS_4_5("claude-opus-4-5")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_5_20251101("claude-opus-4-5-20251101")` Powerful intelligence for long-running agents and coding - `CLAUDE_SONNET_4_5("claude-sonnet-4-5")` **Deprecated**: Will reach end-of-life on November 30, 2026. Please migrate to claude-sonnet-5-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information. High-performance model for agents and coding - `CLAUDE_SONNET_4_5_20250929("claude-sonnet-4-5-20250929")` **Deprecated**: Will reach end-of-life on November 30, 2026. Please migrate to claude-sonnet-5-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information. High-performance model for agents and coding - `class BetaManagedAgentsModelConfigParams` An object that defines additional configuration control over model use - `BetaManagedAgentsModel id` The model that will power your agent. See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. - `Optional effort` How hard Claude works on each inference call. Accepts a bare level string (`"high"`) or `{"type": "high"}`. On create, omitting it resolves the per-model default; on update, omitting it leaves the stored value unchanged. - `enum BetaManagedAgentsEffortLevel` How hard Claude works on each turn. Higher levels favor reasoning depth over latency. Not all models accept every level; invalid combinations are rejected at create time. - `LOW("low")` Low effort. Favors latency over reasoning depth. - `MEDIUM("medium")` Medium effort. Balances latency and reasoning depth. - `HIGH("high")` High effort. Favors reasoning depth. - `XHIGH("xhigh")` Extra-high effort. Not all models accept this level. - `MAX("max")` Maximum effort. Favors reasoning depth over latency. - `class BetaManagedAgentsEffortLow` Low effort. Favors latency over reasoning depth. - `Type type` - `class BetaManagedAgentsEffortMedium` Medium effort. Balances latency and reasoning depth. - `Type type` - `class BetaManagedAgentsEffortHigh` High effort. Favors reasoning depth. - `Type type` - `class BetaManagedAgentsEffortXhigh` Extra-high effort. Not all models accept this level. - `Type type` - `class BetaManagedAgentsEffortMax` Maximum effort. Favors reasoning depth over latency. - `Type type` - `Optional inferenceGeo` Geographic region for model inference. When unset, requests fall through to the workspace's default_inference_geo. On update, `model` is whole-object replacement — omitting inference_geo clears it. - `Optional speed` Inference speed mode. Defaults to `standard`. - `STANDARD("standard")` - `FAST("fast")` - `String name` Human-readable name for the agent. minLength: 1, maxLength: 256 - `Optional description` Description of what the agent does. maxLength: 2048 - `Optional executionIdentity` Execution principal for runs of this agent. Omitting it, sending null, or sending `{"type": "service_account"}` stores the default. - `Optional> mcpServers` MCP servers this agent connects to. Maximum 20. Names must be unique within the array. Every server must be referenced by an `mcp_toolset` in `tools`; unreferenced servers are rejected. See the [MCP connector guide](https://platform.claude.com/docs/en/managed-agents/mcp-connector?sdk_languages=java). - `Type type` - `String name` Unique name for this server, referenced by mcp_toolset configurations. 1-255 characters. minLength: 1, maxLength: 255 - `String url` Endpoint URL for the MCP server. maxLength: 2048 - `Optional metadata` Arbitrary key-value metadata. Maximum 16 pairs, keys up to 64 chars, values up to 512 chars. - `Optional multiagent` Multiagent orchestration configuration. - `Optional> skills` Skills available to the agent. - `class BetaManagedAgentsAnthropicSkillParams` An Anthropic-managed skill. - `Type type` - `String skillId` Identifier of the Anthropic skill (e.g., "xlsx"). minLength: 1, maxLength: 64 - `Optional version` Version to pin. Defaults to latest if omitted. minLength: 1, maxLength: 64 - `class BetaManagedAgentsCustomSkillParams` A user-created custom skill. - `Type type` - `String skillId` Tagged ID of the custom skill (e.g., "skill_01XJ5..."). minLength: 1, maxLength: 64 - `Optional version` Version to pin. Defaults to latest if omitted. minLength: 1, maxLength: 64 - `Optional system` System prompt for the agent. maxLength: 100000 - `Optional> tools` Tool configurations available to the agent. Maximum of 256 tools across all toolsets allowed. - `class BetaManagedAgentsAgentToolset20260401Params` Configuration for built-in agent tools. Use this to enable or disable groups of tools available to the agent. - `Type type` - `Optional> configs` Per-tool configuration overrides. - `class BetaManagedAgentsBashToolConfigParams` Configuration override for the bash tool. - `Optional type` - `JsonValue name = "bash"` Must be "bash". - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `Type type` - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `Type type` - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `JsonValue type = "auto"` - `class BetaManagedAgentsEditToolConfigParams` Configuration override for the edit tool. - `Optional type` - `JsonValue name = "edit"` Must be "edit". - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsReadToolConfigParams` Configuration override for the read tool. - `Optional type` - `JsonValue name = "read"` Must be "read". - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsWriteToolConfigParams` Configuration override for the write tool. - `Optional type` - `JsonValue name = "write"` Must be "write". - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsGlobToolConfigParams` Configuration override for the glob tool. - `Optional type` - `JsonValue name = "glob"` Must be "glob". - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsGrepToolConfigParams` Configuration override for the grep tool. - `Optional type` - `JsonValue name = "grep"` Must be "grep". - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsWebFetchToolConfigParams` Configuration override for the web_fetch tool. - `Optional type` - `JsonValue name = "web_fetch"` Must be "web_fetch". - `Optional> allowedDomains` Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains. - `Optional> blockedDomains` Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains. - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional maxContentTokens` Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs. format: int32 - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `Optional urlSources` Which sources contribute URLs the tool may fetch. Omit to allow every source. - `Optional clientToolResults` Which custom tools' results contribute URLs that may be fetched: "all" (the default), "none", or an only or except list. Each name in a list must be a custom tool in the same tools array. - `enum BetaManagedAgentsWebFetchUrlSourceShorthand` String form of a url_sources value that has no field other than its type: "all" means {"type": "all"} and "none" means {"type": "none"}. - `ALL("all")` - `NONE("none")` - `class BetaManagedAgentsWebFetchUrlSourceToolFilter: union` Which tools' results contribute URLs that may be fetched. - `class BetaManagedAgentsWebFetchUrlSourceAll` Every URL from this source may be fetched. This is the default. - `JsonValue type = "all"` - `class BetaManagedAgentsWebFetchUrlSourceNone` This source contributes no URLs that may be fetched. - `JsonValue type = "none"` - `class BetaManagedAgentsWebFetchUrlSourceOnly` Only the named tools' results contribute URLs that may be fetched. - `JsonValue type = "only"` - `List tools` The tools whose results contribute. Between 1 and 128 entries, each with a different name. An empty list is rejected; use "none" to allow no tool's results. - `JsonValue type = "tool_reference"` Must be "tool_reference". - `String name` Name of the tool. Compared exactly, so upper and lower case letters are different. minLength: 1, maxLength: 128 - `class BetaManagedAgentsWebFetchUrlSourceExcept` Every tool's results contribute URLs that may be fetched, except the named tools' results. - `JsonValue type = "except"` - `List tools` The tools whose results do not contribute. Between 1 and 128 entries, each with a different name. An empty list is rejected; use "all" to leave out no tool's results. - `JsonValue type = "tool_reference"` Must be "tool_reference". - `String name` Name of the tool. Compared exactly, so upper and lower case letters are different. minLength: 1, maxLength: 128 - `Optional serverToolResults` Which of the web_search and web_fetch tools' results contribute URLs that may be fetched: "all" (the default), "none", or an only or except list. Each name in a list must be "web_search" or "web_fetch". - `Optional userInput` Whether URLs in the text of user messages may be fetched: "all" (the default) or "none". - `enum BetaManagedAgentsWebFetchUrlSourceShorthand` String form of a url_sources value that has no field other than its type: "all" means {"type": "all"} and "none" means {"type": "none"}. - `class BetaManagedAgentsWebFetchUrlSourceUserInput: union` Whether URLs in the text of user messages may be fetched. - `class BetaManagedAgentsWebFetchUrlSourceAll` Every URL from this source may be fetched. This is the default. - `class BetaManagedAgentsWebFetchUrlSourceNone` This source contributes no URLs that may be fetched. - `class BetaManagedAgentsWebSearchToolConfigParams` Configuration override for the web_search tool. - `Optional type` - `JsonValue name = "web_search"` Must be "web_search". - `Optional> allowedDomains` Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains. - `Optional> blockedDomains` Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains. - `Optional enabled` Whether this tool is enabled and available to Claude. Overrides the default_config setting. - `Optional permissionPolicy` Permission policy for this tool. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `Optional userLocation` Approximate user location for search result localization. - `JsonValue type = "approximate"` Location precision. Only "approximate" is supported. - `Optional city` City name. minLength: 1, maxLength: 255 - `Optional country` Two-letter ISO 3166-1 country code, uppercase. - `Optional region` Region or state name. minLength: 1, maxLength: 255 - `Optional timezone` IANA timezone identifier, e.g. "America/Los_Angeles". minLength: 1, maxLength: 255 - `Optional defaultConfig` Default configuration applied to all tools in this set. - `Optional enabled` Whether tools are enabled and available to Claude by default. Defaults to true if not specified. - `Optional permissionPolicy` Default permission policy for tools. Controls whether tool calls are auto-approved or require confirmation. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsMcpToolsetParams` Configuration for tools from an MCP server defined in `mcp_servers`. - `Type type` - `String mcpServerName` Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters. minLength: 1, maxLength: 255 - `Optional> configs` Per-tool configuration overrides. - `String name` Name of the MCP tool to configure. 1-128 characters. minLength: 1, maxLength: 128 - `Optional enabled` Whether this tool is enabled. Overrides the `default_config` setting. - `Optional permissionPolicy` Permission policy for this tool. Overrides the `default_config` setting. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `Optional defaultConfig` Default configuration for all tools from this server. - `Optional enabled` Whether tools are enabled by default. Defaults to true if not specified. - `Optional permissionPolicy` Default permission policy for tools from this server. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsCustomToolParams` A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event. - `Type type` - `String description` Description of what the tool does, shown to the agent to help it decide when to use the tool. minLength: 1 - `BetaManagedAgentsCustomToolInputSchema inputSchema` JSON Schema defining the expected input parameters for the tool. - `JsonValue type = "object"` - `Optional properties` - `Optional> required` - `String name` Unique name for the tool. 1-128 characters; letters, digits, underscores, and hyphens. minLength: 1, maxLength: 128 ## Returns - `class BetaManagedAgentsAgent` A Managed Agents `agent`. - `Type type` - `String id` - `Optional archivedAt` When the agent was archived. Null if not archived. format: date-time - `LocalDateTime createdAt` A timestamp in RFC 3339 format format: date-time - `Optional description` - `BetaManagedAgentsExecutionIdentity executionIdentity` Execution principal for runs of this agent. Always present; `{"type": "service_account"}` when runs carry no customer-named principal. - `class BetaManagedAgentsAwsRoleExecutionIdentity` An AWS IAM role principal as returned in API responses. - `JsonValue type = "aws_role"` - `String roleArn` The IAM role's ARN. - `class BetaManagedAgentsServiceAccountExecutionIdentity` The Anthropic-managed service account as returned in API responses. - `JsonValue type = "service_account"` - `List mcpServers` - `Type type` - `String name` - `String url` - `Metadata metadata` - `BetaManagedAgentsModelConfig model` Model identifier and configuration. - `BetaManagedAgentsModel id` The model that will power your agent. See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. - `CLAUDE_HAIKU_5_5("claude-haiku-5-5")` Fastest model for high-volume, real-time tasks - `CLAUDE_SONNET_5_5("claude-sonnet-5-5")` Efficient model for coding and agents - `CLAUDE_OPUS_5_5("claude-opus-5-5")` Powerful intelligence for coding, knowledge work, and long-running agents - `CLAUDE_FABLE_5_1("claude-fable-5-1")` Frontier intelligence for ambitious tasks across coding, scientific discovery, and enterprise workflows - `CLAUDE_SONNET_5("claude-sonnet-5")` Efficient model for coding and agents - `CLAUDE_FABLE_5("claude-fable-5")` Next generation of intelligence for the hardest knowledge work and coding problems - `CLAUDE_OPUS_5("claude-opus-5")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_8("claude-opus-4-8")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_7("claude-opus-4-7")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_6("claude-opus-4-6")` Powerful intelligence for long-running agents and coding - `CLAUDE_SONNET_4_6("claude-sonnet-4-6")` Best combination of speed and intelligence - `CLAUDE_HAIKU_4_5("claude-haiku-4-5")` - `CLAUDE_HAIKU_4_5_20251001("claude-haiku-4-5-20251001")` - `CLAUDE_OPUS_4_5("claude-opus-4-5")` Powerful intelligence for long-running agents and coding - `CLAUDE_OPUS_4_5_20251101("claude-opus-4-5-20251101")` Powerful intelligence for long-running agents and coding - `CLAUDE_SONNET_4_5("claude-sonnet-4-5")` **Deprecated**: Will reach end-of-life on November 30, 2026. Please migrate to claude-sonnet-5-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information. High-performance model for agents and coding - `CLAUDE_SONNET_4_5_20250929("claude-sonnet-4-5-20250929")` **Deprecated**: Will reach end-of-life on November 30, 2026. Please migrate to claude-sonnet-5-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information. High-performance model for agents and coding - `Optional effort` How hard Claude works on each inference call. One of `low`, `medium`, `high`, `xhigh`, `max`. Always present; resolved to the per-model default at save time when not supplied. - `class BetaManagedAgentsEffortLow` Low effort. Favors latency over reasoning depth. - `Type type` - `class BetaManagedAgentsEffortMedium` Medium effort. Balances latency and reasoning depth. - `Type type` - `class BetaManagedAgentsEffortHigh` High effort. Favors reasoning depth. - `Type type` - `class BetaManagedAgentsEffortXhigh` Extra-high effort. Not all models accept this level. - `Type type` - `class BetaManagedAgentsEffortMax` Maximum effort. Favors reasoning depth over latency. - `Type type` - `Optional inferenceGeo` Geographic region for model inference. When unset, requests fall through to the workspace's default_inference_geo. - `Optional speed` Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Defaults to `standard`. Not all models support `fast`; invalid combinations are rejected at create time. - `STANDARD("standard")` - `FAST("fast")` - `Optional multiagent` Multiagent orchestration configuration. Null when the agent is single-threaded. - `class BetaManagedAgentsMultiagentCoordinator` Resolved coordinator topology with a concrete agent roster. - `Type type` - `List agents` Agents the coordinator may spawn as session threads, each resolved to a specific version. - `class BetaManagedAgentsAgentReference` A resolved agent reference with a concrete version. - `Type type` - `String id` - `long version` format: int32 - `class BetaManagedAgentsAdvisor` Platform advisor roster entry: a model the session's primary thread may consult mid-turn. - `Type type` - `String model` The advisor model id. - `class BetaManagedAgentsMultiagent20261001` Resolved multiagent configuration with three members, each enabled or disabled on its own. - `JsonValue type = "multiagent_20261001"` - `BetaManagedAgentsMultiagentAdvisor advisor` Whether the session's primary thread can consult an advisor model. - `class BetaManagedAgentsMultiagentAdvisorEnabled` The session's primary thread can consult `model` mid-turn. - `JsonValue type = "enabled"` - `String model` The advisor model id. - `class BetaManagedAgentsMultiagentAdvisorDisabled` The agent has no advisor. - `JsonValue type = "disabled"` - `BetaManagedAgentsMultiagentSubagents subagents` Whether the agent can spawn session threads. - `class BetaManagedAgentsMultiagentSubagentsEnabled` The agent can spawn session threads. - `JsonValue type = "enabled"` - `BetaManagedAgentsMultiagentInlineAgents inlineAgents` Whether the agent can define inline agents, which are not saved, when it spawns session threads. - `class BetaManagedAgentsMultiagentInlineAgentsEnabled` The agent can define inline agents. - `JsonValue type = "enabled"` - `class BetaManagedAgentsMultiagentInlineAgentsDisabled` The agent cannot define inline agents. - `JsonValue type = "disabled"` - `List predefinedAgents` Predefined agents, which are saved agents that this agent can spawn as session threads, each resolved to a specific version. - `Type type` - `String id` - `long version` format: int32 - `class BetaManagedAgentsMultiagentSubagentsDisabled` The agent cannot spawn session threads. - `JsonValue type = "disabled"` - `BetaManagedAgentsMultiagentWorkflows workflows` Whether the agent can start workflow runs. - `class BetaManagedAgentsMultiagentWorkflowsEnabled` The agent can start workflow runs. - `JsonValue type = "enabled"` - `BetaManagedAgentsMultiagentInlineAgents inlineAgents` Whether a run's plan can define inline agents, which are not saved. - `List predefinedAgents` Predefined agents, which are saved agents that a run's plan can use, each resolved to a specific version. - `Type type` - `String id` - `long version` format: int32 - `class BetaManagedAgentsMultiagentWorkflowsDisabled` The agent cannot start workflow runs. - `JsonValue type = "disabled"` - `String name` - `List skills` - `class BetaManagedAgentsAnthropicSkill` A resolved Anthropic-managed skill. - `Type type` - `String skillId` - `String version` - `class BetaManagedAgentsCustomSkill` A resolved user-created custom skill. - `Type type` - `String skillId` - `String version` - `Optional system` - `List tools` - `class BetaManagedAgentsAgentToolset20260401` - `Type type` - `List configs` - `class BetaManagedAgentsBashToolConfig` Configuration for the bash tool. - `JsonValue type = "bash"` - `boolean enabled` - `JsonValue name = "bash"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `Type type` - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `Type type` - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `JsonValue type = "auto"` - `class BetaManagedAgentsEditToolConfig` Configuration for the edit tool. - `JsonValue type = "edit"` - `boolean enabled` - `JsonValue name = "edit"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsReadToolConfig` Configuration for the read tool. - `JsonValue type = "read"` - `boolean enabled` - `JsonValue name = "read"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsWriteToolConfig` Configuration for the write tool. - `JsonValue type = "write"` - `boolean enabled` - `JsonValue name = "write"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsGlobToolConfig` Configuration for the glob tool. - `JsonValue type = "glob"` - `boolean enabled` - `JsonValue name = "glob"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsGrepToolConfig` Configuration for the grep tool. - `JsonValue type = "grep"` - `boolean enabled` - `JsonValue name = "grep"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsWebFetchToolConfig` Configuration for the web_fetch tool. - `JsonValue type = "web_fetch"` - `boolean enabled` - `JsonValue name = "web_fetch"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `Optional urlSources` Which sources contribute URLs the tool may fetch, always in the object form. Null when not set, which allows every source. - `Optional clientToolResults` Which custom tools' results contribute URLs that may be fetched. Null when not set, which allows every custom tool's results. - `class BetaManagedAgentsWebFetchUrlSourceAll` Every URL from this source may be fetched. This is the default. - `JsonValue type = "all"` - `class BetaManagedAgentsWebFetchUrlSourceNone` This source contributes no URLs that may be fetched. - `JsonValue type = "none"` - `class BetaManagedAgentsWebFetchUrlSourceOnly` Only the named tools' results contribute URLs that may be fetched. - `JsonValue type = "only"` - `List tools` The tools whose results contribute. Between 1 and 128 entries, each with a different name. An empty list is rejected; use "none" to allow no tool's results. - `JsonValue type = "tool_reference"` Must be "tool_reference". - `String name` Name of the tool. Compared exactly, so upper and lower case letters are different. minLength: 1, maxLength: 128 - `class BetaManagedAgentsWebFetchUrlSourceExcept` Every tool's results contribute URLs that may be fetched, except the named tools' results. - `JsonValue type = "except"` - `List tools` The tools whose results do not contribute. Between 1 and 128 entries, each with a different name. An empty list is rejected; use "all" to leave out no tool's results. - `JsonValue type = "tool_reference"` Must be "tool_reference". - `String name` Name of the tool. Compared exactly, so upper and lower case letters are different. minLength: 1, maxLength: 128 - `Optional serverToolResults` Which of the web_search and web_fetch tools' results contribute URLs that may be fetched. Null when not set, which allows both. - `Optional userInput` Whether URLs in the text of user messages may be fetched. Null when not set, which allows them. - `class BetaManagedAgentsWebFetchUrlSourceAll` Every URL from this source may be fetched. This is the default. - `class BetaManagedAgentsWebFetchUrlSourceNone` This source contributes no URLs that may be fetched. - `Optional> allowedDomains` - `Optional> blockedDomains` - `Optional maxContentTokens` format: int32 - `class BetaManagedAgentsWebSearchToolConfig` Configuration for the web_search tool. - `JsonValue type = "web_search"` - `boolean enabled` - `JsonValue name = "web_search"` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `Optional> allowedDomains` - `Optional> blockedDomains` - `Optional userLocation` Approximate user location for search result localization. - `JsonValue type = "approximate"` Location precision. Only "approximate" is supported. - `Optional city` City name. minLength: 1, maxLength: 255 - `Optional country` Two-letter ISO 3166-1 country code, uppercase. - `Optional region` Region or state name. minLength: 1, maxLength: 255 - `Optional timezone` IANA timezone identifier, e.g. "America/Los_Angeles". minLength: 1, maxLength: 255 - `BetaManagedAgentsAgentToolsetDefaultConfig defaultConfig` Resolved default configuration for agent tools. - `boolean enabled` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `class BetaManagedAgentsMcpToolset` - `Type type` - `List configs` - `boolean enabled` - `String name` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `BetaManagedAgentsMcpToolsetDefaultConfig defaultConfig` Resolved default configuration for all tools from an MCP server. - `boolean enabled` - `PermissionPolicy permissionPolicy` Permission policy for tool execution. - `class BetaManagedAgentsAlwaysAllowPolicy` Tool calls are automatically approved without user confirmation. - `class BetaManagedAgentsAlwaysAskPolicy` Tool calls require user confirmation before execution. - `class BetaManagedAgentsAutoPolicy` The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask. - `String mcpServerName` - `class BetaManagedAgentsCustomTool` A custom tool as returned in API responses. - `Type type` - `String description` - `BetaManagedAgentsCustomToolInputSchema inputSchema` JSON Schema for custom tool input parameters. - `JsonValue type = "object"` - `Optional properties` - `Optional> required` - `String name` - `LocalDateTime updatedAt` A timestamp in RFC 3339 format format: date-time - `long version` The agent's current version. Starts at 1 and increments when the agent is modified. format: int32 ## Example ```java package com.anthropic.example; import com.anthropic.client.AnthropicClient; import com.anthropic.client.okhttp.AnthropicOkHttpClient; import com.anthropic.models.beta.agents.AgentCreateParams; import com.anthropic.models.beta.agents.BetaManagedAgentsAgent; import com.anthropic.models.beta.agents.BetaManagedAgentsModel; public final class Main { private Main() {} public static void main(String[] args) { AnthropicClient client = AnthropicOkHttpClient.fromEnv(); AgentCreateParams params = AgentCreateParams.builder() .model(BetaManagedAgentsModel.CLAUDE_OPUS_5_5) .name("My First Agent") .build(); BetaManagedAgentsAgent betaManagedAgentsAgent = client.beta().agents().create(params); } } ``` ### Response (200) ```json { "id": "agent_011CZkYpogX7uDKUyvBTophP", "archived_at": null, "created_at": "2026-03-15T10:00:00Z", "description": "A general-purpose starter agent.", "execution_identity": { "type": "service_account" }, "mcp_servers": [ { "name": "example-mcp", "type": "url", "url": "https://example-server.modelcontextprotocol.io/sse" } ], "metadata": { "foo": "bar" }, "model": { "id": "claude-opus-5-5", "effort": { "type": "low" }, "inference_geo": "inference_geo", "speed": "standard" }, "multiagent": { "advisor": { "type": "disabled" }, "subagents": { "inline_agents": { "type": "enabled" }, "predefined_agents": [ { "id": "agent_011CZkYqphY8vELVzwCUpqiQ", "type": "agent", "version": 1 } ], "type": "enabled" }, "type": "multiagent_20261001", "workflows": { "inline_agents": { "type": "enabled" }, "predefined_agents": [ { "id": "agent_011CZkYqphY8vELVzwCUpqiQ", "type": "agent", "version": 1 } ], "type": "enabled" } }, "name": "My First Agent", "skills": [ { "skill_id": "xlsx", "type": "anthropic", "version": "1" }, { "skill_id": "skill_011CZkZFNu9hAbo3jZPRgTmx", "type": "custom", "version": "2" } ], "system": "You are a general-purpose agent that can research, write code, run commands, and use connected tools to complete the user's task end to end.", "tools": [ { "configs": [ { "enabled": true, "name": "bash", "permission_policy": { "type": "always_allow" }, "type": "bash" } ], "default_config": { "enabled": true, "permission_policy": { "type": "always_ask" } }, "type": "agent_toolset_20260401" } ], "type": "agent", "updated_at": "2026-03-15T10:00:00Z", "version": 1 } ```