> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.levrage.ai/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.levrage.ai/_mcp/server. # Get agent GET https://api.levrage.ai/v1/agents/{agent_id} Get full details of a specific agent by ID. Reference: https://docs.levrage.ai/api-reference/agents/get-agent ## Authentication - `Authorization` header (bearer token, required) — Bearer authentication of the form `Bearer `, where token is your auth token. ## Request ### Path parameters - `agent_id` (string, required) ## Response ### 200 Successful Response - `data` (AgentDetail, required) — Full agent detail response. - `request_id` (string, required) - `timestamp` (datetime, required) - `success` (boolean, optional, default: true) ## Errors ### 422 Unprocessable Entity Error Validation Error - `detail` (list of ValidationError, optional) ## Types ### AgentDetail Full agent detail response. - `id` (string, required) — Agent ID - `name` (string, required) — Agent name - `language` (string, required) — Agent language - `agent_type` (string, required) — Agent type: sales, support, custom - `description` (string, optional, nullable) — Agent description - `custom_description` (string, optional, nullable) — Custom agent description - `sector` (string, optional, nullable) — Industry sector - `tone` (string, optional, nullable) — Agent tone - `agent_role` (string, optional, nullable) — Agent role - `ai_model` (string, optional, nullable) — AI model: lai-premium-2, lai-premium-3, lai-premium-4 - `voice` (string, optional, nullable) — Voice gender (male/female) - `voice_id` (string, optional, nullable) — Specific voice ID for TTS - `accent` (string, optional, nullable) — Voice accent - `pronunciations` (map from string to string, optional, nullable) — Custom pronunciation rules - `voice_response_settings` (VoiceResponseSettings, optional, nullable) — Voice & response settings - `background_sounds_enabled` (boolean, optional, nullable) — Office background sound during calls (volume: voice_response_settings.background_sound_volume) - `prompt` (string, optional, nullable) — Agent instructions/prompt - `start_flow` (string, optional, nullable) — Opening greeting - `end_flow` (string, optional, nullable) — Closing message - `details_to_collect` (list of any, optional, nullable) — Details to collect during call - `brand_name` (string, optional, nullable) — Company/brand name - `typed_knowledge` (string, optional, nullable) — Knowledge base text - `knowledge_base_ids` (list of string, optional) — Knowledge bases the agent searches during calls - `timezone` (string, optional, nullable) — Agent timezone - `other_languages` (list of string, optional, nullable) — Additional languages the agent supports - `webhook_config` (AgentDetailWebhookConfig, optional, nullable) — Webhook configuration - `custom_api` (list of AgentDetailCustomApiItems, optional, nullable) — Custom API tool configs - `lookup_api` (AgentDetailLookupApi, optional, nullable) — Pre-call data lookup config - `transfer_map` (list of AgentDetailTransferMapItems, optional, nullable) — Human handoff numbers - `conversation_flow_id` (string, optional, nullable) — The call flow the agent follows (GET /v1/flows/\{id}), when one is on - `is_active` (boolean, optional, default: true) — Whether agent is active - `created_at` (datetime, optional, nullable) — Creation timestamp - `updated_at` (datetime, optional, nullable) — Last update timestamp ### ValidationError - `loc` (list of ValidationErrorLocItems, required) - `msg` (string, required) - `type` (string, required) ### VoiceResponseSettings Voice detection and response configuration for the agent. - `listening_sensitivity` (double, optional, nullable) — Voice detection sensitivity. 0.1 = very sensitive, 1 = needs loud/clear voice. Default: 0.5 - `wait_before_responding` (integer, optional, nullable) — Padding (ms) before agent starts speaking. Default: 200ms - `silence_to_end_turn` (integer, optional, nullable) — Silence duration (ms) to detect end of user turn. Default: 300ms - `allow_user_to_interrupt` (boolean, optional, nullable) — Allow user to interrupt agent mid-speech. Default: true - `words_to_interrupt` (integer, optional, nullable) — Minimum words before an interruption is accepted. Default: 2 - `min_response_wait` (double, optional, nullable) — Minimum seconds to wait before responding. Default: 0.15 - `max_response_wait` (double, optional, nullable) — Maximum seconds to wait before responding. Default: 0.3 - `fast_response` (boolean, optional, nullable) — Enable preemptive generation for faster responses. Default: true - `voice_detection_sensitivity` (double, optional, nullable) — Voice detection threshold for premium models. Default: 0.5 - `quick_response_padding` (integer, optional, nullable) — Response padding (ms) for realtime models. Default: 200ms - `turn_end_silence` (integer, optional, nullable) — Silence to detect end of turn in realtime models. Default: 300ms - `vad_type` (string, optional, nullable) — VAD type: 'server_vad' (silence-based) or 'semantic_vad' (AI-based). Default: server_vad - `eagerness` (string, optional, nullable) — For semantic_vad: 'low', 'medium', 'high', 'auto'. Default: medium - `remind_on_silence` (boolean, optional, nullable) — Send reminders when user is silent. Default: true - `silence_before_reminder` (integer, optional, nullable) — Seconds of silence before sending reminder. Default: 8 - `max_reminders` (integer, optional, nullable) — Max reminders before auto-disconnect. Default: 5 - `custom_reminder_messages` (list of string, optional, nullable) — Custom reminder messages. E.g. ['Hello, are you there?', 'Please say something...'] - `detect_ivr_systems` (boolean, optional, nullable) — Auto-detect IVR/voicemail systems for outbound calls. Default: true - `enable_dtmf_input` (boolean, optional, nullable) — Enable DTMF keypress detection (press 1 for X). Default: false - `leave_voicemail` (boolean, optional, nullable) — Leave voicemail when answering machine detected. Default: false - `voicemail_message` (string, optional, nullable) — What to say on a voicemail, word for word. Empty = the agent words it from its instructions. - `amd_enabled` (boolean, optional, nullable) — Outbound: listen for an answering machine or phone menu in the first seconds. Default: false - `amd_strategy` (enum, optional, nullable) — How hard to look for a machine: fast, balanced or thorough. Default: balanced - Allowed values: `fast`, `balanced`, `thorough` - `amd_max_wait_seconds` (double, optional, nullable) — Give up looking for a machine after this many seconds. Default: 6 - `amd_voicemail_action` (enum, optional, nullable) — On an answering machine: hangup, leave_message (voicemail_message) or continue. Default: hangup - Allowed values: `hangup`, `leave_message`, `leave_voicemail`, `continue` - `amd_ivr_action` (enum, optional, nullable) — On a phone menu: hangup, press (keys) or continue. Default: hangup - Allowed values: `hangup`, `press`, `navigate`, `continue` - `hold_detection` (boolean, optional, nullable) — When put on hold, wait for a person instead of talking to the hold system. Default: true - `hold_timeout_seconds` (integer, optional, nullable) — Hang up after this long on hold. Default: 90 - `live_call_summary` (boolean, optional, nullable) — On a transfer, brief the person on the call before connecting the caller. Default: false - `max_call_duration_minutes` (VoiceResponseSettingsMaxCallDurationMinutes, optional, nullable) — Auto-end call N minutes after it is answered; fractions allowed (0.5 = 30 seconds). Default: 60 - `max_duration_farewell` (string, optional, nullable) — Said just before the time limit ends the call. Empty = end without a goodbye. - `end_call_mode` (enum, optional, nullable) — auto = the agent hangs up itself when it's done; manual = the call ends when the agent says its closing line. Default: auto - Allowed values: `auto`, `manual` - `greeting_delay` (integer, optional, nullable) — Delay (seconds) before agent greets on outbound calls. Default: 0 - `wait_for_user_response` (boolean, optional, nullable) — Outbound: stay quiet until the person speaks, then greet. Default: false - `answer_wait_timeout` (integer, optional, nullable) — With wait_for_user_response: seconds of silence before the call counts as no answer. Default: 12 - `greeting_allow_interruptions` (boolean, optional, nullable) — The caller can cut the greeting short. Default: true - `call_analysis_enabled` (boolean, optional, nullable) — Enable post-call AI analysis (sentiment, summary). Default: true - `qa_enabled` (boolean, optional, nullable) — Quality audit of each finished call. Default: false - `immediate_transfer` (boolean, optional, nullable) — Immediately transfer call to human without AI interaction. Default: false - `remember_callers` (boolean, optional, nullable) — Inbound: greet a caller from the last 90 days knowing their name and last call. Default: false - `callback_missed_calls` (boolean, optional, nullable) — Inbound: ring back callers whose call was missed. Default: false - `ai_disclosure` (enum, optional, nullable) — Says it is a virtual assistant: if_asked, or at_start (first reply). Unset = its instructions decide. - Allowed values: `if_asked`, `at_start` - `adaptive_interruption` (boolean, optional, nullable) — Keep talking through 'haan', 'hmm', 'ok' while the caller listens. Default: true - `repeat_when_unclear` (boolean, optional, nullable) — Ask the caller to repeat when their audio breaks up. Default: on - `unclear_audio_threshold` (double, optional, nullable) — How unsure (0.3–0.95) before asking to repeat. Default: 0.7 - `noise_cancellation` (boolean, optional, nullable) — Filter background noise from the caller's audio. Default: false - `backchannel_enabled` (boolean, optional, nullable) — Short acknowledgements ('mm-hmm') while the caller talks at length. Default: false - `backchannel_phrases` (list of string, optional, nullable) — The acknowledgements to use. - `voice_speed` (double, optional, nullable) — Speaking speed: 0.8 slower, 1.0 normal, 1.2 faster. Default: 1.0 - `remove_filler_words` (list of string, optional, nullable) — Words dropped from what the caller said before the agent reads it, e.g. ['umm', 'like']. - `background_sound_volume` (double, optional, nullable) — Office background sound volume, 0 to 1 (agent's background_sounds_enabled turns it on). Default: 0.8 - `thinking_sound_volume` (double, optional, nullable) — Keyboard sound while the agent looks something up, 0 to 1. Default: 0.3 - `multilingual` (boolean, optional, nullable) — Reply in whichever of the agent's languages the caller speaks. Off = always the main language. - `language_switch_mode` (enum, optional, nullable) — mirror = follow the caller turn by turn; fixed = ask once, then keep to it. Default: mirror - Allowed values: `mirror`, `fixed` - `number_language` (enum, optional, nullable) — Numbers in English words, or in the language being spoken. Unset = English. - Allowed values: `english`, `native` - `dialects` (map from string to string, optional, nullable) — Regional words per language, e.g. \{'telugu': 'telangana'}. - `natural_code_mixing` (boolean, optional, nullable) — Keep everyday English words (payment, OTP) the way people say them. Default: false - `advanced_intelligence` (boolean, optional, nullable) — Work out what the caller meant before replying. Default: false - `response_creativity` (double, optional, nullable) — Response creativity. 0 = focused, 1 = creative. Default: 0.8 - `turn_detection` (string, optional, nullable) — Turn detection mode: 'vad', 'stt', 'semantic', 'english'. Default: vad - `frontend_actions` (list of VoiceResponseSettingsFrontendActionsItems, optional, nullable) — List of UI actions the agent can trigger on the frontend. - `copilot_mode` (boolean, optional, nullable) — Enable Co-Pilot mode to monitor and intervene in website sessions. - `copilot_critical_events` (list of VoiceResponseSettingsCopilotCriticalEventsItems, optional, nullable) — List of critical user events for Co-Pilot intervention. - `skills` (SkillsConfig, optional, nullable) — Skills the agent is built from (see GET /skills). ### AgentDetailWebhookConfig Webhook configuration ### AgentDetailCustomApiItems ### AgentDetailLookupApi Pre-call data lookup config ### AgentDetailTransferMapItems ### ValidationErrorLocItems - string - integer ### VoiceResponseSettingsMaxCallDurationMinutes Auto-end call N minutes after it is answered; fractions allowed (0.5 = 30 seconds). Default: 60 - integer - double ### VoiceResponseSettingsFrontendActionsItems ### VoiceResponseSettingsCopilotCriticalEventsItems ### SkillsConfig Skills installed on the agent (GET /skills lists them). The agent's instructions are built from these on every call or chat; its own prompt is kept as notes. - `enabled` (boolean, optional, nullable) — Build the agent from its skills. - `installed` (list of string, optional, nullable) — Skill names, the main one first, e.g. ['payment-reminder', 'human-handoff']. - `settings` (map from string to string, optional, nullable) — Each skill's settings by name (GET /skills shows them). - `tools` (map from string to string, optional, nullable) — Your system's URL for a skill step, by tool name, e.g. \{'send\_payment\_link': 'https\://…'}. - `additions` (map from string to SkillsConfigAdditions, optional, nullable) — Your own lines inside a skill, by skill name: \{steps: \{'STEP 3': text}, if\_caller\_says: \[\{says, do}], rules: \[text], collect: \[\{ask, at}]}. - `role` (string, optional, nullable) — Job title the agent introduces itself with. ### SkillsConfigAdditions ## Examples **Response** ```json { "data": { "id": "550e8400-e29b-41d4-a716-446655440000", "name": "Acme Sales Agent", "language": "English", "agent_type": "sales", "description": "Qualifies leads for personal loan products", "custom_description": null, "sector": "Finance & Banking", "tone": "professional", "agent_role": "Sales Agent", "ai_model": "lai-premium-2", "voice": "female", "voice_id": null, "accent": null, "pronunciations": { "Levrage": "Lev-rahj" }, "voice_response_settings": { "listening_sensitivity": 0.5, "wait_before_responding": 200, "silence_to_end_turn": 300, "allow_user_to_interrupt": true, "words_to_interrupt": 2, "min_response_wait": 0.15, "max_response_wait": 0.3, "fast_response": true, "voice_detection_sensitivity": 0.5, "quick_response_padding": 200, "turn_end_silence": 300, "vad_type": "server_vad", "eagerness": "medium", "remind_on_silence": true, "silence_before_reminder": 10, "max_reminders": 3, "custom_reminder_messages": null, "detect_ivr_systems": true, "enable_dtmf_input": false, "leave_voicemail": false, "max_call_duration_minutes": 15, "greeting_delay": 0, "call_analysis_enabled": true, "immediate_transfer": false, "response_creativity": 0.8, "turn_detection": "vad" }, "prompt": "You are an AI sales agent for Acme Finance. Qualify leads for personal loans.", "start_flow": "Hello! I'm calling from Acme Finance. Am I speaking with the right person?", "end_flow": "Thank you for your time! Have a great day.", "details_to_collect": [ { "key": "full_name", "type": "text", "description": "Customer's full name" }, { "key": "monthly_income", "type": "number", "description": "Customer's monthly income" } ], "brand_name": "Acme Finance", "typed_knowledge": "Personal loan products from 50,000 to 25,00,000 INR. Interest rates starting at 10.5% p.a.", "timezone": "Asia/Kolkata", "other_languages": [ "Hindi" ], "webhook_config": { "url": "https://api.yourcompany.com/webhooks/levrage", "events": [ "call.completed", "call.failed" ] }, "custom_api": [], "lookup_api": { "enabled": true, "api_name": "Customer CRM Lookup", "api_url": "https://api.yourcompany.com/customers", "method": "GET", "query_params": [ { "key": "phone", "value": "{number}", "is_variable": true } ], "headers": [ { "key": "Authorization", "value": "Bearer your_crm_api_key", "is_variable": false } ], "timeout_seconds": 10, "retry_count": 1 }, "transfer_map": [ { "name": "Sales Manager", "number": "9876543210", "country_code": "+91", "category": "escalation" } ], "is_active": true, "created_at": "2026-01-15T09:30:00+05:30", "updated_at": "2026-01-15T09:30:00+05:30" }, "request_id": "req_a1b2c3d4e5f6", "timestamp": "2026-01-15T09:30:00+05:30", "success": true } ``` **SDK Code** ```python import requests url = "https://api.levrage.ai/v1/agents/agent_id" headers = {"Authorization": "Bearer "} response = requests.get(url, headers=headers) print(response.json()) ``` ```javascript const url = 'https://api.levrage.ai/v1/agents/agent_id'; const options = {method: 'GET', headers: {Authorization: 'Bearer '}}; try { const response = await fetch(url, options); const data = await response.json(); console.log(data); } catch (error) { console.error(error); } ``` ```go package main import ( "fmt" "net/http" "io" ) func main() { url := "https://api.levrage.ai/v1/agents/agent_id" req, _ := http.NewRequest("GET", url, nil) req.Header.Add("Authorization", "Bearer ") res, _ := http.DefaultClient.Do(req) defer res.Body.Close() body, _ := io.ReadAll(res.Body) fmt.Println(res) fmt.Println(string(body)) } ``` ```ruby require 'uri' require 'net/http' url = URI("https://api.levrage.ai/v1/agents/agent_id") http = Net::HTTP.new(url.host, url.port) http.use_ssl = true request = Net::HTTP::Get.new(url) request["Authorization"] = 'Bearer ' response = http.request(request) puts response.read_body ``` ```java import com.mashape.unirest.http.HttpResponse; import com.mashape.unirest.http.Unirest; HttpResponse response = Unirest.get("https://api.levrage.ai/v1/agents/agent_id") .header("Authorization", "Bearer ") .asString(); ``` ```php request('GET', 'https://api.levrage.ai/v1/agents/agent_id', [ 'headers' => [ 'Authorization' => 'Bearer ', ], ]); echo $response->getBody(); ``` ```csharp using RestSharp; var client = new RestClient("https://api.levrage.ai/v1/agents/agent_id"); var request = new RestRequest(Method.GET); request.AddHeader("Authorization", "Bearer "); IRestResponse response = client.Execute(request); ``` ```swift import Foundation let headers = ["Authorization": "Bearer "] let request = NSMutableURLRequest(url: NSURL(string: "https://api.levrage.ai/v1/agents/agent_id")! as URL, cachePolicy: .useProtocolCachePolicy, timeoutInterval: 10.0) request.httpMethod = "GET" request.allHTTPHeaderFields = headers let session = URLSession.shared let dataTask = session.dataTask(with: request as URLRequest, completionHandler: { (data, response, error) -> Void in if (error != nil) { print(error as Any) } else { let httpResponse = response as? HTTPURLResponse print(httpResponse) } }) dataTask.resume() ```