--- title: "Create An Agent From A Template." method: POST path: "/v1/orchestrate/agents/create-from-template" tags: ["Agents"] --- # Create An Agent From A Template. `POST /v1/orchestrate/agents/create-from-template` Creates a new agent and its dependencies from a template. ## Query parameters - `parent_agent_id` union - string, uuid - string ## Request body - CreateAgentFromTemplate — Schema for creating an agent from template. - `id` string, uuid, nullable — Optional UUID for the agent - `name` string, required — Name of the agent - `category` string, required — Category of the agent - `version` string, nullable — Version of the agent - `description` string, required — Description of what the agent does - `spec` CreateAgent, required — Schema for creating a new agent derived from base. - `name` string, nullable — Name of the agent - `display_name` string, nullable — Display name of the agent - `description` string, required — Description of what the agent does - `instructions` string, nullable — Instructions for the agent - `tools` string[], nullable — List of tool names available to the agent - `collaborators` string[], nullable — List of other agent names this agent can collaborate with - `llm` string, nullable — LLM identifier for the agent - `style` 'default' | 'react' | 'planner' | 'custom' | 'react_intrinsic' | 'experimental_customer_care' | 'code_act', required - `supported_apps` SupportedApp[], nullable — Supported apps of the agent - `glossary` string[], nullable — List of glossary available to the agent - `guidelines` Guideline[], nullable — List of guidelines for the agent behavior - `display_name` string, nullable — Display name of the guideline - `condition` string, required — Condition when the guideline should be applied - `action` string, required — Action to take when the condition is met - `tool` string, nullable — The tool id use for this guideline - `knowledge_base` string[], nullable — List of knowledge_base available to the agent - `hidden` boolean — To show or hide agent when agents are listed - `structured_output` object, nullable — JSON schema defining the structure for agent responses - `custom_join_tool` string, nullable — Reference to a Python tool ID that will be used for custom synthesis of task results - `additional_properties` AgentAdditionalPropertiesIn — A specification for creating settings within an agent. Validation runs for Input. - `starter_prompts` AgentCustomizedPromptsIn - `customize` AgentPrompt[] — list of starter prompts of the agent - `id` string — Unique identifier for the prompt - `title` string, required — Title of the prompt - `subtitle` string — description of the prompt - `prompt` string, required — contents of the prompt - `state` string — active/inactivate - `welcome_content` AgentWelcomeContentIn - `welcome_message` string, nullable — Welcome message - `description` string, nullable — Description - `is_default_voice_greeting` boolean — is the welcome message the default voice greeting - `icon` string — A raw SVG XML string (normalized from bytes / single- / double-escaped) that passes square-icon validation. When serialized, it's just a string. - `realtime_agent_settings` RealtimeAgentSettingsIn — Input model for creating/updating realtime agent settings - `enabled` boolean, nullable — Whether the realtime agent is enabled - `context_settings` ContextSettings — Context management settings for an agent. - `context_compaction_enabled` boolean, nullable — Enable context compaction - `context_compaction_threshold` integer, nullable — Token threshold for context compaction - `compaction_sliding_window` integer, nullable — Sliding window size for compaction - `large_message_threshold` integer, nullable — Token threshold for large message handling - `large_message_chunk_size` integer, nullable — Chunk size for large message processing - `large_message_target_summary` integer, nullable — Target summary size for large messages - `large_message_detect_structured` boolean, nullable — Enable structured data detection - `tags` string[], nullable — List of tags associated to agent - `chat_with_docs` ChatWithDocsConfigInput - `enabled` boolean — Controls if the Chat with Documents feature is enabled for this agent - `supports_full_document` boolean — Controls if the Agent can request to use the full document. Only used if generation.enabled = false - `vector_index` KnowledgeBaseBuiltInVectorIndexConfig - `embeddings_model_name` string, nullable - `chunk_size` integer - `chunk_overlap` integer - `limit` integer - `extraction_strategy` 'express' | 'standard' | 'high_quality' - `generation` GenerationConfiguration — example { "model_id": "meta-llama/llama-3-1-70b-instruct", "prompt_instruction": "When the documents are in different languages, you should respond in english.", "max_docs_passed_to_llm": 5, "retrieval_confidence_threshold": "Lowest", "generated_response_length": "Moderate", "response_confidence_threshold": "Low", "display_text_no_results_found": "no docs found", "display_text_connectivity_issue": "conn failed", } - `model_id` string, nullable - `prompt_instruction` string - `max_docs_passed_to_llm` integer - `generated_response_length` 'Concise' | 'Moderate' | 'Verbose' - `display_text_no_results_found` string - `display_text_connectivity_issue` string - `idk_message` string - `enabled` boolean - `query_rewrite` QueryRewriteConfig — example { "enabled": True, "model_id": "meta-llama/llama-3-1-70b-instruct" } - `enabled` boolean - `model_id` string, nullable - `confidence_thresholds` ConfidenceThresholds - `retrieval_confidence_threshold` 'Off' | 'Lowest' | 'Low' | 'High' | 'Highest' - `response_confidence_threshold` 'Off' | 'Lowest' | 'Low' | 'High' | 'Highest' - `citations` CitationsConfig — example: { "citation_title": "how do i know", "citations_shown": 5, } - `citation_title` string - `citations_shown` integer - `hap_filtering` HAPFiltering — example { "output": { "enabled": True, "threshold": 0.7, } } - `output` HAPFilteringConfig - `enabled` boolean - `threshold` number - `query_source` 'SessionHistory' | 'Agent' - `agent_query_description` string - `context_access_enabled` boolean — Enable access to context variables for this agent - `context_variables` string[], nullable — List of context variable names that this agent can access - `hide_reasoning` boolean — Enable/disable reasoning trace - `voice_configuration_id` string, uuid, nullable — Voice configuration for the agent - `restrictions` 'editable' | 'non_editable' | 'custom' - `bundled_agent_id` string, uuid, nullable — Base agent id - `llm_config` ModelConfigInput - `decoding_method` string, nullable - `prompt` union - string - string[] - `max_tokens` integer, nullable - `max_completion_tokens` integer, nullable - `temperature` number, nullable - `top_p` number, nullable - `n` integer, nullable - `stream` boolean, nullable - `logprobs` integer, nullable - `top_logprobs` boolean, nullable - `echo` boolean, nullable - `stop` union - string - string[] - `presence_penalty` number, nullable - `frequency_penalty` number, nullable - `best_of` integer, nullable - `logit_bias` object, nullable - `user` string, nullable - `context` string, nullable - `examples` Examples[], nullable - `input` unknown, required - `output` unknown, required - `top_k` integer, nullable - `response_format` ResponseFormat - `type` string, required - `json_schema` unknown - `seed` integer, nullable - `store` boolean, nullable - `metadata` object, nullable - `modalities` string[], nullable - `audio` AudioConfig - `voice` string, required - `format` string, required - `service_tier` string, nullable - `prediction` Prediction - `type` string, required - `content` union, required - PredictionContent[] - `type` string, required - `text` string, required - string - `safety_settings` unknown - `anthropic_beta` string, nullable - `anthropic_version` string, nullable - `thinking` Thinking - `type` string, nullable - `budget_tokens` integer, required - `space_id` string, nullable - `project_id` string, nullable - `reasoning_effort` string, nullable - `agent_mapping` object, nullable — agent_config available to the agent - `icon` string — A raw SVG XML string (normalized from bytes / single- / double-escaped) that passes square-icon validation. When serialized, it's just a string. - `dependencies` object — Dependencies including tools and agents ## Response `200` Successful Response - unknown ## Other responses - `422` — Validation Error --- [API](https://skmtc.net/ibm/apis/wxo-server-api.md) · [All operations](https://skmtc.net/ibm/apis/wxo-server-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/ibm/wxo-server-api/versions/82c114c37196/schema)