> ## Documentation Index > Fetch the complete documentation index at: https://docs.talkpilot.io/llms.txt > Use this file to discover all available pages before exploring further. # List performance records > Returns the QA/performance records (`call_statistics`) for the specified agent — latencies, token stats, pipeline configuration and redflags. Results are ordered by `created_at` descending (newest first). The transcript is **not** included here: use `GET /v1/agents/{agentId}/performance/{statisticId}` for a single record with its transcript. Internal debug logs are never returned. Requires the `performance:read` permission. With Dashboard JWT auth this endpoint is restricted to the `super_admin`, `dev_admin` and `dev_employee` roles, mirroring the Dashboard's Performance Analysis page. ## OpenAPI ````yaml get /v1/agents/{agentId}/performance openapi: 3.1.0 info: title: TalkPilot API version: 1.0.0 description: > The TalkPilot API allows you to programmatically manage your call center agents, employees, forwarding slots, knowledge base, and more. ## Authentication All API requests require an API key passed via the `X-API-Key` header: ``` X-API-Key: tp_live_abc123def456... ``` An API key can reach every organization its creator is a member of (see `GET /v1/organization`) and can be restricted to specific agents and permissions. Create and manage keys in the TalkPilot Dashboard under **Settings > API** (Dev-Admin only). ## Rate Limiting Default limits per API key: **60 requests/minute** and **1,000 requests/hour** (both enforced, both configurable per key). Rate limit headers are included in every response to a request authenticated with an API key (Dashboard JWT requests are not rate limited): - `X-RateLimit-Limit` — Max requests per minute - `X-RateLimit-Remaining` — Remaining requests in the current minute window - `X-RateLimit-Reset` — Unix timestamp roughly 60 seconds in the future When rate limited, the API returns `429 Too Many Requests` with `Retry-After: 60` (minute window) or `Retry-After: 3600` (hour window). ## Pagination List endpoints support pagination via query parameters: - `page` — Page number (default: 1) - `limit` — Items per page (default: 20, max: 100) Paginated responses include a `pagination` object with `page`, `limit`, `total`, and `pages`. ## Error Format All errors follow a consistent format: ```json { "error": { "code": "VALIDATION_ERROR", "message": "Human-readable description", "details": [{ "field": "name", "message": "Required" }], "request_id": "req_abc123" } } ``` contact: name: TalkPilot Support email: support@talkpilot.io license: name: Proprietary identifier: LicenseRef-Proprietary servers: - url: https://api.talkpilot.io description: Production security: - apiKey: [] tags: - name: Health description: API health and status - name: Agents description: Manage AI agent configurations - name: Workflow description: Multi-state conversation flow of an agent - name: Tools description: Manage agent tools (HTTP requests, call transfers, etc.) - name: Employees description: Manage employees linked to agents - name: Contacts description: >- Customer contacts (customer_database) per agent — the voice agent recognises callers and greets them by name or tag from here - name: Forwarding Slots description: Manage call forwarding slot assignments - name: Knowledge Base description: Manage agent knowledge base documents - name: Calls description: Read call records and transcripts - name: Performance description: >- Read call performance/QA data (latencies, token stats, redflags, transcript incl. tool calls) - name: Organization description: Organization-level settings - name: API Keys description: Manage API keys (Dashboard JWT auth only; dev_admin) paths: /v1/agents/{agentId}/performance: get: tags: - Performance summary: List performance records description: > Returns the QA/performance records (`call_statistics`) for the specified agent — latencies, token stats, pipeline configuration and redflags. Results are ordered by `created_at` descending (newest first). The transcript is **not** included here: use `GET /v1/agents/{agentId}/performance/{statisticId}` for a single record with its transcript. Internal debug logs are never returned. Requires the `performance:read` permission. With Dashboard JWT auth this endpoint is restricted to the `super_admin`, `dev_admin` and `dev_employee` roles, mirroring the Dashboard's Performance Analysis page. operationId: listPerformanceRecords parameters: - $ref: '#/components/parameters/AgentId' - $ref: '#/components/parameters/PageParam' - $ref: '#/components/parameters/LimitParam' - name: from in: query description: Filter records created after this date (ISO 8601) schema: type: string format: date-time - name: to in: query description: Filter records created before this date (ISO 8601) schema: type: string format: date-time responses: '200': description: List of performance records content: application/json: schema: type: object properties: data: type: array items: $ref: '#/components/schemas/PerformanceRecord' pagination: $ref: '#/components/schemas/Pagination' example: data: - id: 9f1c2d3e-4a5b-6c7d-8e9f-0a1b2c3d4e5f call_id: lk_call_abc123 caller_phone: '+491701234567' agent_phone_number: '+4930123456' agent_id: 550e8400-e29b-41d4-a716-446655440000 duration_seconds: 184 disconnect_reason: user_hangup session_status: completed e2e_latency_p90: 1.42 e2e_latency_median: 0.98 e2e_latency_min: 0.61 transcription_latency_p90: 0.31 transcription_latency_median: 0.22 transcription_latency_min: 0.14 llm_latency_p90: 0.82 llm_latency_median: 0.55 llm_latency_min: 0.33 tts_latency_p90: 0.29 tts_latency_median: 0.21 tts_latency_min: 0.12 llm_total_tokens: 4820 llm_prompt_tokens: 4120 llm_completion_tokens: 700 pre_call_variables: caller_name: Hans Meier extracted_variables: appointment_date: '2026-04-02' llm_provider: openai llm_model: gpt-4o-mini tts_provider: elevenlabs stt: deepgram pipeline_mode: stt_llm_tts created_at: '2026-03-19T10:30:00Z' final_result: outcome: true outcome_label: success final_score: 8 summary_reasoning: Termin erfolgreich vereinbart, keine Rueckfragen offen. prompt_fingerprint: >- 3b1f0c9d8e7a6b5c4d3e2f1a0b9c8d7e6f5a4b3c2d1e0f9a8b7c6d5e4f3a2b1c assistant_misunderstood_multiple_times: detected: false reasoning: Keine Rueckfragen des Assistenten notwendig. score_impact: 0 repeated_same_sentence: detected: false reasoning: Keine Wiederholungen erkannt. score_impact: 0 abrupt_end_without_goodbye: detected: false reasoning: Gespraech wurde regulaer beendet. score_impact: 0 connection_issue_repeated: detected: false reasoning: Stabile Verbindung. score_impact: 0 assistant_policy_violation: detected: false reasoning: Keine Richtlinienverstoesse. score_impact: 0 technical_error: detected: false reasoning: Keine technischen Fehler. score_impact: 0 early_hangup_by_caller: detected: false reasoning: Anrufer blieb bis zum Ende in der Leitung. score_impact: 0 pagination: page: 1 limit: 20 total: 342 pages: 18 '401': $ref: '#/components/responses/Unauthorized' '403': $ref: '#/components/responses/Forbidden' '429': $ref: '#/components/responses/RateLimited' components: parameters: AgentId: name: agentId in: path required: true description: Agent UUID schema: type: string format: uuid PageParam: name: page in: query description: Page number schema: type: integer minimum: 1 default: 1 LimitParam: name: limit in: query description: Items per page schema: type: integer minimum: 1 maximum: 100 default: 20 schemas: PerformanceRecord: type: object description: >- Performance/QA record of one call (list view — no transcript, no internal debug logs) properties: id: type: string format: uuid call_id: type: string description: External call ID (e.g. LiveKit) — links to the `calls` resource caller_phone: type: - string - 'null' agent_phone_number: type: - string - 'null' agent_id: type: - string - 'null' format: uuid duration_seconds: type: integer description: Call duration in seconds disconnect_reason: type: - string - 'null' session_status: type: - string - 'null' e2e_latency_p90: type: - number - 'null' description: End-to-end response latency, 90th percentile (seconds) e2e_latency_median: type: - number - 'null' e2e_latency_min: type: - number - 'null' transcription_latency_p90: type: - number - 'null' description: Speech-to-text latency, 90th percentile (seconds) transcription_latency_median: type: - number - 'null' transcription_latency_min: type: - number - 'null' llm_latency_p90: type: - number - 'null' description: Language model latency, 90th percentile (seconds) llm_latency_median: type: - number - 'null' llm_latency_min: type: - number - 'null' tts_latency_p90: type: - number - 'null' description: Text-to-speech latency, 90th percentile (seconds) tts_latency_median: type: - number - 'null' tts_latency_min: type: - number - 'null' llm_total_tokens: type: - integer - 'null' llm_prompt_tokens: type: - integer - 'null' llm_completion_tokens: type: - integer - 'null' pre_call_variables: type: - object - 'null' additionalProperties: true description: Variables handed to the agent before the call started extracted_variables: type: - object - 'null' additionalProperties: true description: Variables the agent extracted during the call llm_provider: type: - string - 'null' llm_model: type: - string - 'null' tts_provider: type: - string - 'null' stt: type: - string - 'null' description: Speech-to-text provider pipeline_mode: type: - string - 'null' created_at: type: - string - 'null' format: date-time final_result: oneOf: - $ref: '#/components/schemas/PerformanceFinalResult' - type: 'null' prompt_fingerprint: type: - string - 'null' description: >- SHA-256 of the prompt template used — groups calls of the same prompt version assistant_misunderstood_multiple_times: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' repeated_same_sentence: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' abrupt_end_without_goodbye: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' connection_issue_repeated: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' assistant_policy_violation: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' technical_error: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' early_hangup_by_caller: oneOf: - $ref: '#/components/schemas/FailureCause' - type: 'null' Pagination: type: object properties: page: type: integer limit: type: integer total: type: integer pages: type: integer PerformanceFinalResult: type: object description: Overall automated evaluation of the call properties: outcome: type: - boolean - 'null' outcome_label: type: string enum: - success - failed - neutral final_score: type: number description: Overall quality score of the call summary_reasoning: type: string FailureCause: type: object description: | Automated QA assessment of a single failure pattern ("redflag"). Written by the analysis pipeline, not by the caller. properties: detected: type: boolean description: Whether this failure pattern occurred in the call reasoning: type: string description: Short explanation of the assessment score_impact: type: number description: How strongly this pattern lowered the final score Error: type: object required: - error properties: error: type: object required: - code - message - request_id properties: code: type: string enum: - VALIDATION_ERROR - UNAUTHORIZED - FORBIDDEN - AGENT_NOT_ACCESSIBLE - NOT_FOUND - CONFLICT - AGENT_HAS_NUMBER - POOL_EMPTY - NOT_POOL_NUMBER - RATE_LIMITED - INTERNAL_ERROR message: type: string details: type: array items: type: object properties: field: type: string message: type: string request_id: type: string responses: Unauthorized: description: Missing or invalid authentication content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: UNAUTHORIZED message: Invalid or missing API key request_id: req_abc123 Forbidden: description: > Insufficient permissions (`FORBIDDEN`), or the agent is outside the key's `allowed_agent_ids` (`AGENT_NOT_ACCESSIBLE`) content: application/json: schema: $ref: '#/components/schemas/Error' examples: missingPermission: value: error: code: FORBIDDEN message: 'API key lacks required permission: agents:write' request_id: req_abc123 agentNotAccessible: value: error: code: AGENT_NOT_ACCESSIBLE message: API key does not have access to this agent request_id: req_abc123 RateLimited: description: Too many requests for this API key (60/minute or 1,000/hour by default) headers: Retry-After: description: >- Seconds to wait before retrying — 60 for the minute window, 3600 for the hour window schema: type: integer X-RateLimit-Limit: schema: type: integer X-RateLimit-Remaining: schema: type: integer X-RateLimit-Reset: schema: type: integer content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: RATE_LIMITED message: Rate limit exceeded. Try again in 60 seconds. request_id: req_abc123 securitySchemes: apiKey: type: apiKey in: header name: X-API-Key description: | API key for external access. Format: `tp_live_<32-hex-chars>`. Create keys in the TalkPilot Dashboard under Settings > API. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.