Connectors

Agno AgentOS GET /metrics (self-hosted base_url). Authority: apis.io agno.

Setup documented

TENG-2205: Direct AI21 Labs API calls (Jamba models via the ai21 SDK, no framework) called from customer code. AI21 has no admin or billing API — confirmed against AI21's own docs (docs.ai21.com/docs/usage-cost): usage and billing are visible only under the Studio console (Account > Model usage, Account > Billing & Plans), never through a programmatic endpoint. This connector only receives data if the customer's own code emits an OTel span or log record per chat call carrying the attributes below; an uninstrumented integration reports nothing, not zero. OpenAI-compatibility of AI21's chat API could not be independently confirmed in this research pass (AI21's API-reference subpages were unreachable from this environment); treat it as unconfirmed rather than assumed. There is no telara-observability SDK wrapper for AI21. Instrumentation must be manual: either hand-written OTel spans around the ai21 SDK call site, or — if AI21's endpoint does turn out to be OpenAI-compatible — a generic OpenAI SDK wrapper pointed at AI21's base URL, emitting the gen_ai.* attributes below.

Setup documented

Aider AI pair-programming CLI — no hosted SaaS and no Aider-operated REST API (apis.io provider text + synthetic CLI OpenAPI stubs under /commands/*). Governance path: force traffic through an LLM gateway (LiteLLM / Portkey / OpenRouter / Helicone) with IdP-issued keys, or wrap the OpenAI-compatible client with telara-observability[openai] when using a custom base_url. This connector documents the OTLP instrument path for wrapped clients.

Setup documented

Airtable is a low-code platform to build collaborative apps using a spreadsheet-database hybrid

Setup documented

Amazon Q Developer — account-level adoption metrics (invocations, active users, generated lines) via CloudWatch. Spend connector for the AI Platform Spend Governance feature.

Setup documented

Amazon Redshift data warehouse service

Setup documented

Amazon Web Services (AWS) is a comprehensive cloud computing platform providing infrastructure services including compute, storage, databases, networking, and security.

Setup documented

Amplitude product analytics for dashboards, events, cohorts, annotations, experiments, and release-impact analysis.

Setup documented

Anthropic organization Admin API — workspace, member, API key management plus Usage & Cost reporting for per-workspace / per-key token spend governance.

Setup documented

Direct Anthropic SDK calls (no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

Apollo.io B2B data and sales engagement. People and organization search, enrichment by email or domain, sequences, and email tracking.

Setup documented

Argo CD GitOps deployment platform for applications, sync status, projects, clusters, repositories, and release drift.

Setup documented

Arize Phoenix / OpenInference OTel GenAI → Telara OTLP receiver.

Setup documented

Asana is a work management platform where teams can stay connected and organized

Setup documented

Ashby recruiting platform for jobs, candidates, applications, interviews, offers, and hiring analytics.

Setup documented

Augment coding assistant — no usage export on apis.io; OTLP/gateway path.

Setup documented

Auth0 identity platform for users, organizations, applications, roles, connections, and tenant audit logs.

Setup documented

AutoGen / AG2 multi-agent telemetry ingested via OpenTelemetry. AG2's native instrumentation (instrument_agent / instrument_llm_wrapper / instrument_pattern) emits GenAI-convention spans for every conversation turn, LLM call, and tool execution — point its exporter at the Telara OTLP receiver or use the telara-observability shim.

Setup documented

AWS Bedrock — account-level token + invocation metrics via CloudWatch, daily cost via Cost Explorer. Spend connector for the AI Platform Spend Governance feature.

Setup documented

AWS Strands Agents framework — OTLP spans via telara-observability. Captures per-agent-run Bedrock/Claude token consumption and cost. Use the telara-observability shim (`instrument(framework="aws-strands", ...)`) or configure OTEL_EXPORTER_OTLP_ENDPOINT to point at the Telara OTLP receiver. Cost is resolved per the underlying provider model (Bedrock or Anthropic direct). Requires pip extra: strands.

Setup documented

Microsoft Azure is a cloud computing platform providing infrastructure, platform, and software services including virtual machines, storage, and managed databases.

Setup documented

Azure DevOps engineering system for projects, repositories, pipelines, builds, releases, work items, and engineering delivery signals.

Setup documented

Azure OpenAI — per-request token usage pulled from the tenant's Log Analytics workspace (diagnostic settings). The only major cloud exposing provider-side per-request token logs. Spend-only connector for the AI Platform Spend Governance feature.

Setup documented

Direct Azure OpenAI SDK calls (no framework) instrumented by the telara-observability provider wrapper. Complements the azure_openai pull connector (provider-side TokenUsageLogs) with client-side per-run spans.

Setup documented

TENG-2205: Direct Baidu ERNIE (Wenxin Yiyan) API calls via the Qianfan platform, called from customer code. Baidu has no admin or billing API — usage and spend are console-only, in the Qianfan console (console.bce.baidu.com/qianfan). This connector only receives data if the customer's own code emits an OTel span or log record per chat call carrying the attributes below; an uninstrumented integration reports nothing, not zero. LOW CONFIDENCE, stated plainly per TENG-2205 instructions rather than guessed around: Qianfan's documentation is almost entirely Chinese-language, and every attempt to reach Baidu's docs/console domains from this environment this pass returned a connection failure (ECONNREFUSED), so neither the exact base URL nor OpenAI-compatibility could be independently verified live. OpenRouter's catalog does corroborate ERNIE 4.5 model variants exist as an active lab offering, which is the only independent confirmation obtained this pass. OpenAI-compatibility is UNCONFIRMED. Do not assume a generic OpenAI SDK wrapper works against Qianfan's endpoint without first checking Baidu's (Chinese-language) API reference directly. There is no telara-observability SDK wrapper for Baidu/Qianfan. Instrumentation must be manual: hand-written OTel spans around the qianfan SDK call site, emitting the gen_ai.* attributes below. The `spend.currency: USD` below is an ASSUMPTION, not a confirmed fact — Qianfan's console pricing may be denominated in CNY. Confirm against Baidu's live pricing page before populating baidu_ernie.model_price_table with real numbers.

Setup documented

BambooHR HRIS for employee directory, employment status, time off, job info, and lightweight people operations workflows.

Setup documented

Basecamp is a project management tool that keeps teams organized and on track

Setup documented

Baseten OpenAI-compatible / HTTP inference — telara-observability OTLP Guide.

Setup documented

Bitbucket code collaboration platform

Setup documented

Braintrust AI evaluation and tracing — pull per-experiment and per-span cost and usage from the Braintrust API. Connects spend governance to AI evaluation workflows.

Setup documented

Braze customer engagement platform for campaigns, canvases, segments, message analytics, and lifecycle growth workflows.

Setup documented

Brevo transactional email (SMTP API) — send emails, inspect templates, review delivery statistics/events, manage transactional-contact and domain blocklists, and read inbound-parsing events.

Setup documented

Brex corporate cards, expenses, transactions, and reimbursements. Webhook-driven for real-time receipt filing and policy enforcement.

Setup documented

Bria enterprise/IP-safe visual generation API (docs.bria.ai — image generation, background removal, image/video editing endpoints, trained on licensed data) called directly from customer code. TENG-2205: no telara-observability SDK wrapper exists for Bria, and there is no OpenAI-compatible fallback — Bria exposes several distinct operation types (generate, background-remove, video-edit, ...) with their own request/ response shapes, not a single chat-completion call. This connector only receives data if the customer's own code emits an OTel span or log record per operation call carrying the attributes below; an uninstrumented integration reports nothing, not zero. Bria does NOT bill in credits — its public pay-as-you-go pricing is a fixed $ rate per operation (e.g. ~$0.03/image generated, ~$0.018/ background removal, ~$0.0225-0.045/second for video edits), but the rate differs by operation type and there is no admin/billing API to confirm which rate applied to a given call, which is why this is otlp_push rather than a pull connector. Cost is NOT derived to USD here: because the per-unit rate depends on which of several operation types was called and Bria exposes no reliable operation-type tag on the response, hardcoding a single model_price_table would silently misprice a subset of calls — worse than no dollar figure. This connector reports operation and video-second counts only; $ appears only if the customer's own code computes and emits a cost_usd attribute using its own knowledge of which endpoint it called.

Setup documented

Buildkite CI/CD for organizations, pipelines, builds, jobs, agents, annotations, and deployment evidence.

Setup documented

Cal.com is an open-source scheduling platform for booking meetings, managing event types, and coordinating availability.

Setup documented

Capsule CRM platform for managing contacts and organizations (parties), sales opportunities, tasks, and pipelines.

Setup documented

Cerebras OpenAI-compatible / HTTP inference — telara-observability OTLP Guide.

Setup documented

Cerner is an electronic health records (EHR) system providing healthcare IT solutions

Setup documented

Chargebee subscription billing and revenue management platform for customers, subscriptions, and invoices.

Setup documented

ChatGPT Enterprise / Edu — OpenAI Compliance Logs Platform (Audit + Auth + Codex + Conversation logs) for per-user prompt counts, tool usage, GPT invocations, and SSO-anchored governance. Spend-only connector for the AI Platform Spend Governance feature.

Setup documented

CircleCI API v2 for projects, pipelines, workflows, jobs, artifacts, and CI/CD run evidence.

Setup documented

Civitai image-generation API (civitai.com — community model hub with a generation API for its hosted Stable Diffusion / Flux-family models) called directly from customer code. TENG-2205: no telara-observability SDK wrapper exists for Civitai, and there is no OpenAI-compatible fallback — Civitai's generation endpoints take model/workflow-specific parameters, not a chat-completion shape. This connector only receives data if the customer's own code emits an OTel span or log record per generate call carrying the attributes below; an uninstrumented integration reports nothing, not zero. Civitai meters usage in "Buzz" — its own platform credit, consumed per generation at a rate that varies by model and resolution. Civitai already offers a native OAuth (authorization code + PKCE) app model with per-app Buzz spend caps enforced platform-side, so for OAuth-integrated apps Civitai itself is the primary spend control; this OTLP connector is a customer-side, per-user usage view on top of that — not a replacement for it. There is no admin/billing API exposing historical Buzz consumption by user, which is why this is otlp_push rather than a pull connector. Cost is NOT derived to USD: Buzz is sold in bundles at a blended rate that varies by purchase tier and is not a stable public $/Buzz figure, so no model_price_table is set. This connector reports Buzz-credit and call counts only; $ appears only if the customer's own code emits a cost_usd attribute.

Setup documented

Claude Code (Anthropic CLI) telemetry ingested via OpenTelemetry. Customers set CLAUDE_CODE_ENABLE_TELEMETRY=1 and point OTEL_EXPORTER_OTLP_ENDPOINT at mcp-gateway's OTLP receiver. Each LLM request span / log record is feeded through the same engine pipeline (billing → ratelimit → compliance → audit) as MCP tool calls. Reference: https://code.claude.com/docs/en/monitoring-usage

Setup documented

Claude for Enterprise — Compliance API (Activity Feed) + Enterprise Analytics API for per-user / per-workspace usage, cost, and audit governance. Spend-only connector for the AI Platform Spend Governance feature.

Setup documented

Cleo is an EDI and integration platform for B2B communications and supply chain management

Setup documented

ClickUp is a project management platform that brings tasks, docs, goals, and chat into a single app.

Setup documented

Cline (and Cline Enterprise) telemetry ingested via OpenTelemetry. Customer configures Cline's OTel exporter to point at mcp-gateway's OTLP receiver. Token+cost data lands on the `task.tokens` event; broader productivity signals (tasks completed, tool usage) on the other task.* events. Reference: https://docs.cline.bot/enterprise-solutions/monitoring/opentelemetry Event reference: https://docs.cline.bot/enterprise-solutions/monitoring/opentelemetry-events

Setup documented

Clio is a cloud-based legal practice management software for law firms

Setup documented

Close CRM platform for managing leads, contacts, sales opportunities, tasks, and activity notes.

Setup documented

Cloudflare API for DNS, zones, Workers, Pages, WAF, audit logs, and Zero Trust Access/Tunnel operations.

Setup documented

Coda docs and tables for product specs, operating cadences, planning trackers, and lightweight internal apps.

Setup documented

Direct Cohere API calls (cohere SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). pip extra: pip install telara-observability[cohere]

Setup documented

Direct calls to the CometAPI router's OpenAI-compatible API (base_url=https://api.cometapi.com/v1), instrumented via the OpenAI SDK wrapper — no first-party CometAPI wrapper exists in telara-observability, but CometAPI documents itself as "OpenAI Compatible" so existing OpenAI-SDK code (and its observability wrapper) works unmodified against the router's base URL. pip extra: pip install telara-observability[openai] CometAPI aggregates 500+ models from multiple upstream providers behind one key and offers its own dashboard (spend, latency, call volume, budget alerts) but no bulk usage/billing pull API — dashboard only, same limitation as the other routers in this batch. CometAPI sits between the customer and the underlying model vendor, so cost attribution lands on CometAPI as the router, not on the origin model provider — a real chargeback limitation: the origin vendor's own admin-API pull (if connected) will not show these calls.

Setup documented

Confluence team collaboration and documentation platform

Setup documented

Continue.dev open-source AI coding assistant — OTLP telemetry via environment variables. Captures per-request token usage across any configured LLM provider (OpenAI, Anthropic, Ollama, Azure, etc.). No pip install needed: set OTel env vars in your shell or CI environment and Continue.dev emits spans to the Telara OTLP receiver automatically. # Environment variable setup (no code change required): # OTEL_EXPORTER_OTLP_ENDPOINT=<TELARA_OTLP_ENDPOINT> # OTEL_EXPORTER_OTLP_HEADERS="Authorization=Bearer <API_KEY>" # OTEL_SERVICE_NAME=continue-dev

Setup documented

CrewAI crew telemetry ingested via OpenTelemetry. CrewAI ships native OTel spans (GenAI semantic conventions, SafeOTLPSpanExporter) — point the exporter at the Telara OTLP receiver, or use the telara-observability shim (`instrument(framework="crewai", ...)`) for run/user stamping. One span per agent step / task / LLM call with crew + agent attribution.

Setup documented

CrewAI AMP/Cloud — OTLP from crews (Bearer org tokens are control-plane only).

Setup documented

CrowdStrike Falcon endpoint security for hosts, detections, alerts, incidents, sensor health, and threat response.

Setup documented

Cursor AI coding assistant — Admin + Analytics APIs for per-user spend, AI requests, tab acceptance, and team usage governance.

Setup documented

Customer.io lifecycle messaging and growth automation for people, events, campaigns, newsletters, broadcasts, and journeys.

Setup documented

Databricks is a unified analytics platform for big data and AI

Setup documented

Datadog monitoring and analytics platform

Setup documented

Deepgram GET /v1/projects/{project_id}/usage. Speech spend connector.

Setup documented

Direct DeepInfra API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.deepinfra.com/v1/openai. DeepInfra has no admin/billing API — this OTLP path is the only way to attribute spend, since the OpenAIInstrumentor patches the openai SDK client itself regardless of base_url. pip extra: pip install telara-observability[openai]

Setup documented

Direct DeepSeek API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.deepseek.com. pip extra: pip install telara-observability[openai]

Setup documented

Devin (Cognition) autonomous software engineer — Consumption API for per-session ACU usage, outcomes, and org usage metrics. Spend-only connector for the AI Platform Spend Governance feature.

Setup documented

Dify LLM app platform — self-hosted / Cloud workflows emit OpenTelemetry when configured (or via Langfuse/OTEL plugins). Point the OTLP exporter at Telara for per-run agent/workflow token and cost governance. apis.io lists Chat / Completion / Conversations / Datasets / Files / Workflows APIs only — no usage or billing export — so Connect stays an OTLP Guide.

Setup documented

Discord team communication platform

Setup documented

DocuSeal open-source e-signature platform for document templates, signature-request submissions, submitters, and signed-document retrieval.

Setup documented

DocuSign eSignature for envelopes, templates, recipients, documents, users, and agreement workflow evidence.

Setup documented

Drata compliance automation for controls, evidence, tests, personnel, vendors, risks, and audit workflows.

Setup documented

Dropbox Sign (formerly HelloSign) e-signature platform for sending, tracking, and templating legally binding signature requests.

Setup documented

Dust.tt workspace analytics — GET /api/v1/w/{wId}/analytics/export (tables: users, usage_metrics, active_users, agents, messages, skill_usage, tool_usage, source). Deprecated sibling: workspace-usage (remove after 2026-06-01 per OpenAPI). Spend connector for AI Platform Spend Governance.

Setup documented

Direct Eden AI API calls (edenai SDK or raw HTTP, no framework). Eden AI is a multi-provider aggregator — "one API to access OpenAI, Mistral, Gemini, Claude, and dozens of specialized AI providers" for chat, embeddings, translation and more — at base URL https://api.edenai.run/v2/ (confirmed partially live: Eden AI's own docs state "Both share the same base URL (https://api." before the fetched excerpt was truncated; the domain matches the well-documented api.edenai.run). Eden AI also publishes an OpenAI-compatible endpoint ("drop-in replacement — change the base URL"), and per its pricing docs charges "no markup on provider pricing — you pay exactly what the underlying provider charges, plus a [platform fee]", so cost is provider-rate-based even though it is billed through Eden AI. Because Eden AI fronts chat/completion models as well as embeddings, an output_tokens metric is meaningful here and is kept. IMPORTANT for chargeback: the OTLP spans this connector ingests carry Eden AI's own request/response envelope. Whichever underlying provider actually served the call, cost and usage land on Eden AI as the billing entity in this pipeline, not on the underlying model vendor directly — a real limitation for per-vendor chargeback. Per-underlying-provider attribution is only possible if the customer's own instrumentation additionally captures the provider/model Eden AI reports back (gen_ai.response.model), which this connector does try via model_fallback_paths. Eden AI has no separate admin/billing API beyond its dashboard, so spend is only capturable via this customer-side OTLP push. No telara-observability provider wrapper exists for Eden AI today (the package's optional-dependencies only cover langchain/langgraph/crewai/ autogen/llamaindex/openai-agents/openai/anthropic/azure-openai) — spans must come from the customer's own OTel instrumentation around the Eden AI SDK/ HTTP call, or from the OpenAI instrumentation package if they call Eden AI through its OpenAI-compatible endpoint. Spans should be tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

Elasticsearch is a distributed, RESTful search and analytics engine capable of addressing a growing number of use cases

Setup documented

Epic is an electronic health records (EHR) system used by healthcare organizations

Setup documented

Factory (Droids) — Enterprise Analytics API for Factory Standard Credits, per-user token spend, tool invocations, and productivity signals. Spend-only connector for AI Platform Spend Governance.

Setup documented

Figma is a collaborative interface design tool

Setup documented

Direct Fireworks AI API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.fireworks.ai/inference/v1. pip extra: pip install telara-observability[openai]

Setup documented

Flowise — self-hosted API key; OTel/Langfuse export → Telara (Dify-class).

Setup documented

Freshservice IT service management for tickets, requesters, agents, assets, changes, problems, and service desk automation.

Setup documented

FriendliAI GET /v1/team/usage + /v1/team/cost. Direct LLM inference spend.

Setup documented

Gainsight customer success platform for companies, relationships, health signals, success plans, and renewal-risk workflows.

Setup documented

Direct calls to a Gcore Everywhere Inference deployment's OpenAI-compatible API (per-deployment base_url, e.g. https://model-<deployment>-<project>-<account>.ai.gcore.dev/v1 — Gcore issues one endpoint per deployed model rather than a single shared router URL), instrumented via the OpenAI SDK wrapper. No first-party Gcore wrapper exists in telara-observability; Gcore's own docs confirm the deployed-model endpoint is queried with the Python/JS OpenAI SDK, so the generic OpenAI wrapper pointed at the deployment's base URL captures spans the same way it does for other OpenAI-compatible endpoints. pip extra: pip install telara-observability[openai] Gcore hosts open-weight models from its Application Catalog (Llama, Mistral, etc.) on Gcore-operated GPU inference nodes and bills per minute of compute time, not per token — so token-derived model_price_table cost here is a best-effort proxy, not Gcore's actual invoiced figure. Gcore sits between the customer and the underlying open-weight model, so cost attribution lands on Gcore as the inference host, not on a separate upstream model vendor (there is none to bill — the weights are open and Gcore is the only biller).

Setup documented

Gemini Code Assist usage and output-signal ingestion via Google Cloud Logging. Reads metadata-log entries (lines accepted, suggestions shown, chat exposures) from a customer-designated centralized logging project, joins identity via protoPayload.authenticationInfo.principalEmail, and emits per-user output metrics for the spend-vs-output scatter. Spend is derived per-seat from GCP billing, not from log events.

Setup documented

Gemini for Google Workspace (formerly Duet AI) — per-user adoption via Google Workspace Admin Reports API. Tracks daily active users and feature usage (Docs, Gmail, Meet AI features) for AI seat governance.

Setup documented

GitHub code collaboration and version control platform

Setup documented

GitHub Actions workflow automation for workflows, workflow runs, jobs, artifacts, logs, and deployment execution evidence.

Setup documented

GitHub Copilot — usage metrics + seat assignment for AI platform spend governance. Distinct from the GitHub SCM connector; this governs Copilot spend and developer-productivity signals only.

Setup documented

GitHub pull-request output-signal ingestion for the AI Platform Spend Governance feature. Counts PRs merged/opened per user across an org's repos to power the spend-vs-output scatter (plan §11.7). NOT a spend source; data_kind=output_signals isolates it from cost-allocation pipelines.

Setup documented

GitLab code collaboration and DevOps platform

Setup documented

GitLab Duo AI metrics — per-user usage across Code Suggestions, Duo Chat, Code Review, Agent Platform, Job Troubleshooting, and MCP tool calls, plus the AI Catalog agents/flows configured for a group. Pairs with `gitlab` connector for code/MR operations.

Setup documented

Glean — Insights API exposing per-user activity (searches, AI assistant usage) across the enterprise knowledge workspace. Spend connector for the AI Platform Spend Governance feature.

Setup documented

Gong revenue intelligence for calls, transcripts, scorecards, users, and sales conversation analytics.

Setup documented

Google Agent Development Kit (ADK) — OTLP spans emitted via the ADK's built-in OpenTelemetry integration. Captures per-agent-run Gemini token usage, tool invocations, and latency. Point OTEL_EXPORTER_OTLP_ENDPOINT at the Telara OTLP receiver, or use the telara-observability shim (`instrument(framework="google-adk", ...)`). Requires pip extra: google-adk.

Setup documented

Google's serverless data warehouse

Setup documented

Google Cloud Identity Groups API for security groups, direct memberships, transitive memberships, and identity-mapped groups.

Setup documented

Google Cloud Platform (GCP) is a suite of cloud computing services providing infrastructure, data analytics, machine learning, and application services.

Setup documented

Direct Google Gemini API calls (google-generativeai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). pip extra: pip install telara-observability[gemini]

Setup documented

Direct Google Vertex AI API calls (google-cloud-aiplatform SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). pip extra: pip install telara-observability[google-cloud]

Setup documented

Google Workspace (G Suite) productivity suite including Gmail, Drive, Sheets, Docs, Calendar, Chat, and read-only Admin Directory context

Setup documented

Grafana is an open source analytics and monitoring solution

Setup documented

Grammarly Business — Analytics API exposing per-user writing + Grammarly-AI usage statistics, plus License Management for seats. Spend connector for the AI Platform Spend Governance feature.

Setup documented

Granola meeting notes and engineering memory platform for summaries, transcripts, attendees, folders, decisions, and action items.

Setup documented

Greenhouse Recruiting Harvest API for jobs, candidates, applications, users, departments, offices, and hiring workflow review.

Setup documented

Direct Groq API calls (groq SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). pip extra: pip install telara-observability[groq]

Setup documented

Harness software delivery platform for pipelines, projects, executions, services, environments, deployments, and feature/release governance.

Setup documented

Harvey (legal AI) — admin Usage History API exposing per-event usage with user attribution across products, workflows, and vault projects. Spend connector for the AI Platform Spend Governance feature.

Setup documented

deepset Haystack (open-source) pipelines/agents — OpenTelemetryTracer / OpenTelemetryConnector (opentelemetry-haystack) emits GenAI spans to any OTLP backend. Point the exporter at Telara for per-pipeline token and cost governance. apis.io also lists haystack-ai (deepset Cloud) with workspaces/pipelines management APIs only — no usage/spend export — so Cloud stays OTLP/gateway, not a pull Connect card.

Setup documented

HCP Terraform and Terraform Enterprise for organizations, projects, workspaces, runs, state versions, and policy evidence.

Setup documented

Helicone LLM proxy and observability — pull per-request cost, latency, and model usage from the Helicone API. Captures usage logged through the Helicone proxy without additional instrumentation.

Setup documented

Real-time webhook delivery of Helicone LLM proxy request events. Complements the `helicone` pull connector (polled /v1/request/query) for tenants that enable Helicone's webhook push instead of, or in addition to, polling. Helicone signs every delivery with HMAC-SHA256 in the helicone-signature header. TENG-2189.

Setup documented

HiBob HRIS for employee lifecycle, people data, time off, documents, reports, and HR-driven access workflows.

Setup documented

Self-hosted Higress AI Gateway spend tracking via customer-side instrumentation. Higress (github.com/alibaba/higress) is an open-source (Apache-2.0), Envoy/Istio-based cloud-native API gateway that originated at Alibaba and is now a CNCF Sandbox project; deployed by the customer via Docker or Helm, never embedded by Telara. Its AI-proxy plugin does OpenAI protocol conversion in front of 100+ model providers, so a customer's OpenAI-compatible client pointed at their own Higress base_url (there is no single shared URL — customers supply their own) is instrumented the same way as any OpenAI-compatible endpoint. No first-party Higress wrapper exists in telara-observability; use the generic OpenAI wrapper against the customer's Higress base_url. pip extra: pip install telara-observability[openai] Because Higress is self-hosted with no Telara-visible vendor invoice, there is no bulk usage/billing API to pull at all — OTLP push from customer instrumentation is the only signal Telara can see. Higress sits between the customer and whichever underlying model vendor it proxies to, so cost attribution lands on the gateway hop as instrumented, not on the origin model provider — a real chargeback limitation, and here also an availability limitation: if the customer does not instrument, there is nothing to attribute.

Setup documented

HubSpot CRM and marketing platform

Setup documented

Hugging Face OpenAI-compatible / HTTP inference — telara-observability OTLP Guide.

Setup documented

Direct Hyperbolic API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.hyperbolic.xyz/v1. Hyperbolic has no admin/billing API (console only) — this OTLP path is the only way to attribute spend, since the OpenAIInstrumentor patches the openai SDK client itself regardless of base_url. pip extra: pip install telara-observability[openai]

Setup documented

Ideogram image-generation API (api.ideogram.ai, Api-Key header) called directly from customer code. TENG-2205: no telara-observability SDK wrapper exists for Ideogram, and there is no OpenAI-compatible fallback a generic wrapper could use instead — image generation is a different call shape (prompt in, image binary/URL out) than a chat completion. This connector only receives data if the customer's own code emits an OTel span or log record per generate call carrying the attributes below; an uninstrumented integration reports nothing, not zero. Ideogram meters usage in CREDITS, and per its docs "credit usage scales with the number of output images and selected model/render speed" — it is not token-based, so gen_ai.usage.* attributes do not apply and are intentionally absent. Credit balance and spend are visible only on the Ideogram account dashboard (ideogram.ai/manage-api); there is no admin/ billing API to pull them, which is why this is otlp_push rather than a pull connector. Cost is NOT derived to USD: Ideogram publishes no public credit-to-dollar rate that holds across plans, so no model_price_table is set. This connector reports credit and image counts only; $ appears only if the customer's own code emits a cost_usd attribute.

Setup documented

incident.io incident management for incidents, alerts, follow-ups, catalog entries, severities, and response workflows.

Setup documented

TENG-2205: Direct Inflection AI API calls (Inflection 3 Pi / Productivity) called from customer code. Inflection has no admin or billing API — console-only. This connector only receives data if the customer's own code emits an OTel span or log record per chat call carrying the attributes below; an uninstrumented integration reports nothing, not zero. First-party API status is UNCLEAR and stated as such rather than guessed: developers.inflection.ai did not resolve (DNS NXDOMAIN, checked twice this pass) and inflection.ai's own site has no visible developer/API section — it is entirely focused on the Pi consumer product. The only independently confirmed access path this pass is via third-party resellers (OpenRouter lists "Inflection 3 Productivity" and "Inflection 3 Pi" through its own unified API), which is consistent with the prior research note that resellers, not Inflection itself, are the better-documented OpenAI-compatible surface. Do not assume a first-party Inflection endpoint exists or is OpenAI-compatible without re-confirming; if a customer is only reachable via a reseller, that reseller's own OTel/OpenAI-compatible wrapper is what actually instruments this call, not anything Inflection ships. There is no telara-observability SDK wrapper for Inflection (first-party or reseller). Instrumentation must be manual: hand-written OTel spans, or an OpenAI-compatible wrapper pointed at whatever base_url the customer is actually calling (first-party or reseller), emitting the gen_ai.* attributes below.

Setup documented

Intercom customer messaging platform for managing contacts, conversations, and support

Setup documented

Jamf Pro device management for Apple computers, mobile devices, inventory, users, groups, and endpoint lifecycle workflows.

Setup documented

JazzHR (formerly The Resumator) applicant tracking system for jobs, applicants, hiring-pipeline stages, recruiting users, activity history, and notes.

Setup documented

Jenkins automation server for jobs, builds, queues, nodes, artifacts, and CI/CD execution evidence.

Setup documented

Direct Jina AI API calls (jina SDK or raw HTTP, no framework). Embeddings live at https://api.jina.ai/v1/embeddings and reranking at https://api.jina.ai/v1/rerank. Jina's embeddings endpoint is deliberately OpenAI-compatible — the vendor states it matches the input/output JSON schema of OpenAI's text-embedding-3-large model, so a customer already emitting gen_ai.* spans for an OpenAI-shaped call gets the same shape here for free. Jina has no admin/billing API (usage is dashboard-only under "API Key & Billing"), so spend is only capturable via this customer-side OTLP push. Token accounting is input-only: embeddings and rerank calls have no completion step, so there is no meaningful output_tokens metric — one that could only ever read zero would be worse than omitting it. input_tokens and total_tokens are what this connector reports instead. No telara-observability provider wrapper exists for Jina today (the package's optional-dependencies only cover langchain/langgraph/crewai/ autogen/llamaindex/openai-agents/openai/anthropic/azure-openai) — spans must come from the customer's own OTel instrumentation around the Jina SDK/HTTP call, or from the OpenAI instrumentation package if they call Jina through an OpenAI-compatible client. Spans should be tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

Jira project management and issue tracking platform

Setup documented

Apache Kafka is an open-source distributed event streaming platform

Setup documented

Kandji Apple device management for devices, blueprints, users, app inventory, and endpoint compliance workflows.

Setup documented

Klaviyo email marketing and customer data platform (CDP) for managing profiles (contacts), lists, segments, campaigns, and flows.

Setup documented

Kubecost Kubernetes cost monitoring for allocation, assets, cloud cost, efficiency, savings, and cluster spend analysis.

Setup documented

Kubernetes is an open-source container-orchestration system for automating deployment, scaling, and management

Setup documented

LangChain agent telemetry ingested via OpenTelemetry. Apps either enable LangChain's native OTel export (LANGSMITH_OTEL_ENABLED + OTEL_EXPORTER_OTLP_ENDPOINT pointed at the Telara OTLP receiver) or use the telara-observability shim (`instrument(framework="langchain", ...)`), which stamps telara.run.id / telara.framework / gen_ai.* on every LLM, tool, and chain span. Each span is one per-event spend record with run grouping, latency, and outcome.

Setup documented

Langdock usage export (POST /export/users). Enterprise AI workspace spend. Authority: apis.io langdock OpenAPI — Bearer only (no public OAuth).

Setup documented

Langflow — customer API key (UI OAuth is Langflow's own login, not Telara).

Setup documented

Langfuse LLM observability — pull per-trace spend, latency, and model usage via the Langfuse public API. Zero-code alternative to OTLP instrumentation for teams already tracing to Langfuse.

Setup documented

LangGraph agent telemetry ingested via OpenTelemetry. LangGraph runs on LangChain's callback system — the same native OTel export or the telara-observability shim applies, plus per-node spans and the graph thread_id as the session grouping key.

Setup documented

LangSmith observability — per-run traces for LangChain/LangGraph agents (latency, tokens, cost, status). Pull connector for teams already tracing to LangSmith; teams without LangSmith use the Telara OTLP instrumentation lane instead.

Setup documented

LangSmith Alerting Rules webhook — push delivery of the runs that triggered an alert rule, in real time, instead of Telara polling /api/v1/runs/query. Complements the `langsmith` pull connector. One delivery covers ALL runs that tripped the rule in its evaluation window ({rule_id, start_time, end_time, runs: [...]}); the receiver splits `runs` into one SpendEvent per run (see mcp-gateway/internal/spendhook/handler.go's decodeEventRows) so per-run granularity is real, not a lie on top of a batched aggregate. LangSmith has no HMAC signing for alert webhooks; the tenant provisions a shared secret via the ?token= query parameter instead. TENG-2189.

Setup documented

LaunchDarkly feature management for projects, environments, feature flags, segments, members, teams, and audit events.

Setup documented

Lever recruiting platform for opportunities, postings, applications, interviews, users, requisitions, and hiring pipeline review.

Setup documented

Linear project management and issue tracking platform

Setup documented

LinkedIn professional network. Profile read, member share posts (UGC), and (with appropriate developer program access) people/organization search and ad accounts.

Setup documented

Self-hosted LiteLLM Proxy spend tracking — pulls /spend/logs for per-request cost, tokens, and user_id attribution. Customers set the proxy base URL as tenant config; Telara never embeds a LiteLLM instance.

Setup documented

Push-based LiteLLM Proxy spend delivery: the proxy's success callback (litellm_settings.success_callback: ["webhook"] / generic HTTP callback) POSTs a top-level JSON ARRAY of StandardLoggingPayload rows directly, instead of Telara polling /spend/logs. Complements the `litellm` pull connector. LiteLLM has no HMAC signing for custom webhooks; the tenant provisions a shared secret via the ?token= query parameter instead. TENG-2189.

Setup documented

LlamaIndex pipeline telemetry ingested via OpenTelemetry. The official opentelemetry-instrumentation-llamaindex package (or the telara-observability shim) traces LLM, embedding, retriever, and agent events to the Telara OTLP receiver.

Setup documented

Mailchimp email marketing platform for managing audiences (lists), campaigns, and list members.

Setup documented

Mailgun transactional email delivery — send messages, and inspect sending domains, mailing lists, and deliverability stats/events.

Setup documented

Make.com — no confirmed per-user AI spend export for Connect yet. Coming soon.

Setup documented

Marketo is a marketing automation platform for lead management, campaigns, and email marketing

Setup documented

Mastra TypeScript agent framework — OtelExporter sends GenAI-convention traces/logs to any OTLP collector. Point the exporter at the Telara OTLP receiver for per-run agent/LLM/tool spend governance. Self-hosted Mastra servers also expose GET /api/telemetry/traces (apis.io Telemetry API) for local pull; Telara Connect uses the OTLP push path so attribution lands in the shared receiver without per-tenant Mastra base URLs.

Setup documented

Mattermost is an open source, self-hosted Slack-alternative from Mattermost, Inc.

Setup documented

Microsoft 365 productivity suite via Microsoft Graph — OneDrive, SharePoint sites, and files (delegated OAuth).

Setup documented

Microsoft 365 Copilot usage via Microsoft Graph Reports API — per-user daily activity (chat queries, document interactions, meeting summaries) for M365 Copilot seat governance.

Setup documented

Microsoft Entra ID identity and governance data for users, groups, service principals, app role mappings, access packages, assignments, policies, and reviews.

Setup documented

Microsoft Teams collaboration platform

Setup documented

Miro collaborative whiteboarding for boards, frames, sticky notes, diagrams, product planning, and engineering discovery workshops.

Setup documented

Direct Mistral AI API calls (mistralai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). pip extra: pip install telara-observability[mistral]

Setup documented

Direct Mixedbread AI API calls (mixedbread-ai SDK or raw HTTP, no framework). All requests go to https://api.mixedbread.com/ (confirmed live against Mixedbread's own docs) — the vendor's published surface is embeddings and reranking, reached under /v1 (per public API reference; the exact path names and usage-field layout could not be independently re-verified live in this session — the deeper API-reference pages 404'd/ 403'd on fetch — so treat the JSONPaths below as the standard gen_ai.* convention, not vendor-confirmed field names). Mixedbread's embeddings route is commonly used as an OpenAI-SDK drop-in (same request/response shape as OpenAI's /v1/embeddings). Mixedbread has no admin/billing API (usage is dashboard-only), so spend is only capturable via this customer-side OTLP push. Token accounting is input-only: embeddings and rerank calls have no completion step, so there is no meaningful output_tokens metric — one that could only ever read zero would be worse than omitting it. input_tokens and total_tokens are what this connector reports instead. No telara-observability provider wrapper exists for Mixedbread today (the package's optional-dependencies only cover langchain/langgraph/ crewai/autogen/llamaindex/openai-agents/openai/anthropic/azure-openai) — spans must come from the customer's own OTel instrumentation around the Mixedbread SDK/HTTP call. Spans should be tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

Mixpanel product analytics for events, funnels, insights, retention, cohorts, and product-growth reporting.

Setup documented

Direct Modular MAX serving-endpoint calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Modular exposes an OpenAI-compatible /v1/chat/completions route both for self-hosted MAX serve (customer's own endpoint, e.g. base_url=http://<host>:8000/v1) and for Modular Cloud's hosted Shared/Dedicated Endpoints — base_url is deployment-specific and set by the customer's own OpenAI SDK client config. Modular has no admin/billing API for either deployment mode (console only) — this OTLP path is the only way to attribute spend, since the OpenAIInstrumentor patches the openai SDK client itself regardless of base_url. pip extra: pip install telara-observability[openai]

Setup documented

Document-oriented NoSQL database

Setup documented

Popular open source relational database

Setup documented

n8n workflow automation. Primary Telara path: native OpenTelemetry traces (Settings → OpenTelemetry / N8N_OTEL_* env) for per-run workflow and AI-node spans. apis.io also lists a Public REST Insights API (GET /insights/summary on https://app.n8n.cloud/api/v1 or {base}/api/v1) — org-aggregate only, no per-user attribution in the published OpenAPI, so it is not the Connect pull surface (OTLP stays the Guide).

Setup documented

Netlify web deployment and hosting platform

Setup documented

NetSuite SuiteTalk REST records for customers, vendors, employees, invoices, sales orders, and finance operations evidence.

Setup documented

New Relic application performance monitoring platform

Setup documented

Direct NLP Cloud API calls (nlpcloud SDK or raw HTTP, no framework). NLP Cloud is a multi-provider aggregator: one API surfaces a catalog of open-source and proprietary text-generation/embedding models behind a single key, at https://api.nlpcloud.io/v1/<model>/<endpoint> (well- documented publicly; NLP Cloud's own docs pages returned 403 to programmatic fetch during this session, so the base URL and per-model path shape could not be independently re-verified live here — treat the JSONPaths below as the standard gen_ai.* convention, not a vendor-confirmed field layout). Unlike Voyage/Jina/Mixedbread, NLP Cloud fronts chat/completion-style models as well as embeddings, so an output_tokens metric is meaningful here and is kept. IMPORTANT for chargeback: because NLP Cloud is an aggregator, the OTLP spans this connector ingests only carry NLP Cloud's own request/response — whichever underlying model actually served the call, cost and usage land on NLP Cloud as the billing entity, not on the underlying model vendor. Per- underlying-model attribution is only possible if the customer's own instrumentation additionally captures the model id NLP Cloud reports back (gen_ai.response.model), which this connector does try via model_fallback_paths. Auth is NOT bearer-prefixed: NLP Cloud expects `Authorization: Token <api_key>` (not `Bearer <api_key>`), reflected below via header_prefix. NLP Cloud has no admin/billing API of its own beyond the dashboard, so spend is only capturable via this customer-side OTLP push. No telara-observability provider wrapper exists for NLP Cloud today (the package's optional-dependencies only cover langchain/langgraph/ crewai/autogen/llamaindex/openai-agents/openai/anthropic/azure-openai) — spans must come from the customer's own OTel instrumentation around the NLP Cloud SDK/HTTP call. Spans should be tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

Notion is an all-in-one workspace for notes, databases, wikis, and project management

Setup documented

Direct Nscale serverless-inference API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://inference.api.nscale.com/v1. Nscale has no admin/billing API (console only) — this OTLP path is the only way to attribute spend, since the OpenAIInstrumentor patches the openai SDK client itself regardless of base_url. pip extra: pip install telara-observability[openai]

Setup documented

Okta identity and access management for users, groups, applications, factors, policies, and system log investigations.

Setup documented

Okta Identity Governance access certification, entitlement, grant, collection, and catalog data for access-review-aligned Telara setup proposals.

Setup documented

Self-hosted local inference via Ollama (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). No cloud cost — tracks token volume and latency only. Uses the OpenAI SDK with base_url=http://localhost:11434/v1. pip extra: pip install telara-observability[openai]

Setup documented

OpenAI organization Admin API — Usage, Costs, and Audit Logs for per-project / per-user / per-key token spend governance.

Setup documented

OpenAI Agents SDK telemetry ingested via OpenTelemetry. The SDK's traces have NO public export API — runs become visible only through a trace processor. Use opentelemetry-instrumentation-openai-agents-v2 or the telara-observability shim (`instrument(framework="openai_agents")`) to convert SDK traces to GenAI-convention spans aimed at the Telara OTLP receiver.

Setup documented

OpenAI Codex (agentic coding) — Codex Analytics API for per-user credits/tokens and Compliance API for audit logs. Spend-only connector for the AI Platform Spend Governance feature.

Setup documented

Direct OpenAI SDK calls (no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

OpenHands — gateway-wrap or instrument model calls → Telara OTLP.

Setup documented

OpenRouter LLM gateway — organization activity and analytics via Management API key (not an inference key). Pulls per-user / per-key / per-model usage and USD spend for AI Platform Spend Governance.

Setup documented

Direct OpenRouter OpenAI-compatible API calls instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so cost rollups exclude them when a framework adapter already covers the same call. Complements the openrouter admin pull connector with client-side per-call spans. Uses the OpenAI SDK with base_url=https://openrouter.ai/api/v1. pip extra: pip install telara-observability[openai]

Setup documented

Opsgenie alerting and incident management — alerts, incidents, on-call schedules, escalations, teams, and heartbeat monitors.

Setup documented

Outreach sales engagement platform for prospects, accounts, sequences, mailings, opportunities, and outbound workflow automation.

Setup documented

Paddle merchant-of-record payments platform for customers, products, prices, subscriptions, transactions, and discounts.

Setup documented

PagerDuty incident management and on-call platform

Setup documented

PandaDoc e-signature and document workflow platform for creating documents from templates, sending them for signature, and tracking signing status.

Setup documented

Pendo product experience platform for visitors, accounts, product analytics, feedback, guides, and in-app adoption programs.

Setup documented

Direct Perplexity AI API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.perplexity.ai. pip extra: pip install telara-observability[openai]

Setup documented

PingOne identity management for environments, populations, users, groups, applications, and group memberships.

Setup documented

Pipedream Connect usage records. Customer creates an OAuth client in their Pipedream workspace (client_credentials at /v1/oauth/token) — not a Telara-registered public OAuth app. Paste client_id, client_secret, project_id.

Setup documented

Pipedrive CRM platform for managing sales deals, contacts (persons and organizations), and pipeline activities.

Setup documented

Plandex — spend via gateway-wrapped / instrumented model calls → OTLP.

Setup documented

PlayHT text-to-speech API (api.play.ht) called directly from customer code. TENG-2205: there is no telara-observability SDK wrapper for PlayHT, and unlike the LLM providers there is no OpenAI-compatible fallback either — a generic chat-completion wrapper cannot be pointed at a TTS synthesize call. This connector only receives data if the customer's own code emits an OTel span or log record per synthesize request carrying the attributes below; an uninstrumented integration reports nothing, not zero. PlayHT meters usage by CHARACTERS submitted for synthesis (its plans are priced by monthly character/word volume), not tokens, so gen_ai.usage.* token attributes do not apply and are intentionally absent here — an always-zero input_tokens field would look like real data and isn't. Cost is NOT derived: PlayHT's public pricing is a tiered monthly character allowance, not a flat per-character $ rate, so no model_price_table is set. Dollar figures only appear if the customer's instrumentation independently emits a cost_usd attribute; absent that, this connector reports character and audio-second counts only.

Setup documented

Portkey AI gateway — Admin / analytics API for per-request cost, tokens, and metadata-based user attribution (`_user`). Pull connector for customers who already standardize traffic through Portkey.

Setup documented

PostgreSQL relational database management system

Setup documented

PostHog product engineering platform for analytics, feature flags, experiments, session replay, surveys, events, and release-impact investigation.

Setup documented

Transactional email delivery via Postmark. Send agent report notifications, HITL approval emails, and other system messages. Each Postmark "server" gets its own credential and outbound activity log.

Setup documented

Microsoft Power BI business analytics service

Setup documented

Prometheus monitoring system and time series database

Setup documented

PromptLayer — OTLP ingest (POST /otlp/v1/traces). Point export at Telara.

Setup documented

PydanticAI agent framework — OTLP spans via telara-observability. Per-run cost, latency, model, and output captured for each PydanticAI agent run. Use the telara-observability shim (`instrument(framework="pydantic-ai", ...)`) or configure PydanticAI's built-in OpenTelemetry export and point OTEL_EXPORTER_OTLP_ENDPOINT at the Telara OTLP receiver. Requires pip extra: pydantic-ai.

Setup documented

QuickBooks Online accounting platform for company info, customers, vendors, invoices, bills, payments, accounts, and financial review workflows.

Setup documented

Recurly subscription billing and revenue management platform for accounts, plans, subscriptions, invoices, transactions, and coupons.

Setup documented

Redis key-value store

Catalog facts only

Replicate OpenAI-compatible / HTTP inference — telara-observability OTLP Guide.

Setup documented

Replit Enterprise — engagement and spend analytics exist in the admin dashboard with SCIM provisioning. apis.io lists Deployments / Repls / Users management APIs only (no usage, credits, or spend export). Catalogued as coming_soon until Replit ships a pull surface; prefer routing model traffic through an LLM gateway (OpenRouter / LiteLLM / Portkey) for attributable spend today.

Setup documented

Direct calls to the Requesty router's OpenAI-compatible API (base_url=https://router.requesty.ai/v1), instrumented via the OpenAI SDK wrapper — no first-party Requesty wrapper exists in telara-observability today, but because Requesty is OpenAI-compatible by design ("One API. 300+ models."), the generic OpenAI wrapper pointed at the router's base URL captures spans the same way it does for any OpenAI-compatible endpoint. pip extra: pip install telara-observability[openai] Requesty returns a `cost` field inline in the `usage` object on every non-streaming response — a cost source most routers do not expose — so cost_paths checks for it before falling back to model_price_table derivation. Requesty sits between the customer and the underlying model vendor (OpenAI, Anthropic, etc.), so cost attribution here lands on Requesty as the router, not on the origin model provider — a real chargeback limitation: the origin vendor's own admin-API pull (if connected) will not show these calls.

Setup documented

Rippling workforce platform for employees, groups, devices, applications, departments, and identity/workforce operations.

Setup documented

SailPoint Identity Security Cloud identity governance data for identities, sources, access profiles, roles, entitlements, and account/source mappings.

Setup documented

Salesflare intelligent CRM for small B2B teams — accounts (companies), contacts, opportunities/deals, tasks, and sales pipeline tracking.

Setup documented

Salesforce CRM platform for managing accounts, contacts, leads, opportunities, and cases

Setup documented

Salesloft sales engagement for people, accounts, cadences, cadence memberships, calls, tasks, and sales activity.

Setup documented

SambaNova OpenAI-compatible / HTTP inference — telara-observability OTLP Guide.

Setup documented

Sana Enterprise (Workday) AI workspace that hosts Telara as a remote MCP server so authorized workspace users can invoke governed Telara tools. Host-only; Telara holds no Sana credential.

Setup documented

Saviynt Enterprise Identity Cloud identity governance data for users, endpoints, connections, SAV roles, and role membership.

Setup documented

Twilio Segment CDP public API for workspaces, sources, destinations, warehouses, tracking plans, and data pipeline governance.

Setup documented

Segmind multi-model image/video generation API (api.segmind.com; hundreds of model endpoints — SDXL, Flux, video and upscaling pipelines, etc.) called directly from customer code. TENG-2205: no telara-observability SDK wrapper exists for Segmind, and there is no OpenAI-compatible fallback — each Segmind model is its own REST endpoint with a model-specific request/response shape, not a chat completion. This connector only receives data if the customer's own code emits an OTel span or log record per inference call carrying the attributes below; an uninstrumented integration reports nothing, not zero. Segmind meters usage in "billing units" (Segmind's own per-model credit system — the unit cost differs per model/pipeline and per output resolution), not tokens, so gen_ai.usage.* attributes do not apply and are intentionally absent. There is no admin/billing API exposing balance or per-call unit cost outside the account dashboard, which is why this is otlp_push rather than a pull connector. Cost is NOT derived to USD: Segmind's billing-unit rate varies per model and is not published as a single stable table, so no model_price_table is set. This connector reports billing-unit and call counts only; $ appears only if the customer's own code emits a cost_usd attribute.

Setup documented

Microsoft Semantic Kernel — OTLP spans via the Semantic Kernel built-in telemetry (Python or .NET). Per-invocation token usage and cost across any underlying LLM provider. Point OTEL_EXPORTER_OTLP_ENDPOINT at the Telara OTLP receiver, or use the telara-observability shim (`instrument(framework="semantic-kernel", ...)`). Python: pip install semantic-kernel .NET: dotnet add package OpenTelemetry (see .NET alternative comment below) # Python instrumentation: # from telara_observability import instrument # instrument(framework="semantic-kernel", service_name="semantic-kernel") # # .NET alternative (no pip install): # services.AddOpenTelemetry() # .WithTracing(b => b.AddOtlpExporter(o => { # o.Endpoint = new Uri("<TELARA_OTLP_ENDPOINT>"); # o.Headers = "Authorization=Bearer <API_KEY>"; # }));

Setup documented

Semgrep AppSec Platform for deployments, projects, findings, rules, scans, and software security evidence.

Setup documented

SendGrid transactional email delivery platform. Send emails, review suppression lists (bounces, blocks, spam reports) for deliverability visibility, inspect transactional templates, and review email activity and send statistics.

Setup documented

Sentry application monitoring and error tracking platform

Setup documented

ServiceNow IT service management platform for incidents, changes, problems, CMDB, and service catalog

Setup documented

Shopify e-commerce platform for managing products, orders, customers, and inventory

Setup documented

Sierra Agent OS — no public usage/admin OpenAPI on apis.io yet. Coming soon.

Setup documented

Direct SiliconFlow API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.siliconflow.com/v1. SiliconFlow has no admin/billing API (console only) — this OTLP path is the only way to attribute spend, since the OpenAIInstrumentor patches the openai SDK client itself regardless of base_url. pip extra: pip install telara-observability[openai]

Setup documented

Slack team communication and collaboration platform

Setup documented

Submit candidate applications and check status on SmartRecruiters career-site job postings. Covers posting setup and apply flows — not full ATS hiring (requisitions, interviews, offers).

Setup documented

HuggingFace Smolagents — OTLP instrumentation via telara-observability. Captures per-step and per-run token usage and cost across models. Use the telara-observability shim (`instrument(framework="smolagents", ...)`) or configure the OTLP exporter directly and point OTEL_EXPORTER_OTLP_ENDPOINT at the Telara OTLP receiver. Requires pip extra: smolagents.

Setup documented

Cloud data platform

Setup documented

Snyk security platform for organizations, projects, targets, issues, dependencies, packages, and vulnerability evidence.

Setup documented

SonarQube code quality and security platform for projects, issues, quality gates, measures, and engineering risk signals.

Setup documented

Sourcegraph Cody / Amp — Analytics API (CSV) for per-user-per-day code AI usage (searches, chats, autocomplete, CAR%). Spend connector for the AI Platform Spend Governance feature.

Setup documented

Splunk log management and analytics platform

Setup documented

Square payments platform for processing card payments, managing orders, customers, product catalog items, and business locations.

Setup documented

Atlassian Statuspage incident and status-page management for pages, components, component groups, incidents (including scheduled maintenance), and subscribers.

Setup documented

TENG-2205: Direct StepFun API calls (api.stepfun.ai/v1/chat/completions, OpenAI-style request/response shape per StepFun's own docs) called from customer code. StepFun has no admin or billing API — console-only. This connector only receives data if the customer's own code emits an OTel span or log record per chat call carrying the attributes below; an uninstrumented integration reports nothing, not zero. NOT independently re-verified live in this research pass: this environment could not reach StepFun's docs (platform.stepfun.com / www.stepfun.com refused connections outright), so the base URL and OpenAI-compatibility claims above rest on prior research, not a fresh fetch. OpenRouter's catalog does corroborate StepFun as an active foundation-model lab (Step3 listed), which is the only independent confirmation obtained this pass. Because StepFun's OpenAI-style shape is plausible (unconfirmed live), a generic OpenAI SDK wrapper pointed at StepFun's base_url may be able to instrument it — but there is no telara-observability SDK wrapper for StepFun specifically, so this still requires either manual OTel spans or a customer-supplied OpenAI-compatible wrapper. IMPORTANT — StepFun meters usage in platform CREDITS (~7,000,000 credits ≈ $1 per prior research), not USD. cost_derivation below resolves to a credit figure via a StepFun-specific price table, NOT a dollar amount; currency is intentionally left unset rather than USD so a raw credit count is never mislabeled as a dollar cost. Do not read the `spend` block's derived value as USD without an explicit credits-to-USD conversion applied downstream.

Setup documented

Stripe payment processing platform for charges, customers, subscriptions, and invoicing

Setup documented

Supabase open-source backend-as-a-service platform

Setup documented

Tabnine Enterprise — admin Usage Metrics API for per-user, per-model token + cost across code completions, chat, agent, and instruct. Spend connector for the AI Platform Spend Governance feature.

Setup documented

Direct Together AI API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.together.xyz/v1. pip extra: pip install telara-observability[openai]

Setup documented

Trello is a visual collaboration tool that creates a shared perspective on any project

Setup documented

Direct calls to the Unify AI router's OpenAI-compatible API (historically base_url=https://api.unify.ai/v0), instrumented via the OpenAI SDK wrapper — no first-party Unify wrapper exists in telara-observability, but Unify's router was documented as OpenAI-compatible, so the generic OpenAI wrapper pointed at the router's base URL is expected to capture spans the same way it does for other OpenAI-compatible endpoints. pip extra: pip install telara-observability[openai] UNVERIFIED AS OF 2026-08: unify.ai's current public site and docs (docs.unify.ai) present a "virtual AI employee" agent product with no visible LLM-router API reference, and the historical router endpoint (api.unify.ai/v0) now 404s. This connector is written against the previously-documented router shape and should be confirmed against a live Unify account before being marked available to customers — the product may have pivoted away from standalone LLM routing. Unify sits between the customer and the underlying model vendor, so cost attribution here lands on Unify as the router, not on the origin model provider — a real chargeback limitation.

Setup documented

Vanta trust management for controls, evidence, tests, personnel, vendors, risks, and audit-readiness workflows.

Setup documented

Vantage FinOps platform for cloud cost reports, saved views, provider accounts, budgets, and engineering spend investigations.

Setup documented

Vapi voice agents — POST /analytics (apis.io). Bearer private key.

Setup documented

Vercel frontend deployment and hosting platform

Setup documented

Direct Voyage AI API calls (voyageai SDK or raw HTTP, no framework). Base URL is https://api.voyageai.com/v1; the vendor exposes POST /embeddings and POST /rerank — embeddings and rerank only, no chat/completion endpoint. No OpenAI-compatible route is published. Voyage has no admin/billing API (usage is dashboard-only), so spend is only capturable via this customer-side OTLP push. Token accounting is input-only: an embedding or rerank call has no completion, so there is no meaningful output_tokens metric — one that could only ever read zero would be worse than omitting it. input_tokens and total_tokens (Voyage's usage object reports both, and they are equal for embedding calls) are what this connector reports instead. No telara-observability provider wrapper exists for Voyage today (the package's optional-dependencies only cover langchain/langgraph/crewai/ autogen/llamaindex/openai-agents/openai/anthropic/azure-openai) — spans must come from the customer's own OTel instrumentation around the Voyage SDK call. Spans should be tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard).

Setup documented

Webflow visual CMS and website platform for managing sites, CMS collections and items, pages, forms, webhooks, and assets.

Setup documented

W&B Weave POST /cost/query (HTTP Basic with API key). Authority: apis.io wandb.

Setup documented

Windsurf (Codeium / Cognition) — User Page Analytics API for per-user activity + prompt-credit usage. Spend connector for the AI Platform Spend Governance feature.

Setup documented

Workable applicant tracking system (ATS) for recruiting — jobs (requisitions) and candidates across the hiring pipeline.

Setup documented

Workday HCM and finance system for workers, organizations, jobs, requisitions, time off, and HR-driven access workflows.

Setup documented

TENG-2205: Direct Writer API calls (api.writer.com, POST /v1/chat) called from customer code. Writer has no admin or billing API — usage and spend are visible only in Writer's own console, so this connector is the only way to attribute Writer spend in Telara, and it only receives data if the customer's own code emits an OTel span or log record per chat call carrying the attributes below; an uninstrumented integration reports nothing, not zero. Writer's request/response schema is proprietary, not OpenAI-compatible: responses carry Writer-specific fields (graph_data, llm_data, translation_data, web_search_data) with no documented OpenAI compatibility layer, so a generic OpenAI SDK wrapper cannot be pointed at it. There is also no telara-observability SDK wrapper for Writer. Instrumentation must be manual: either hand-written OTel spans around the writer-sdk call site, or a customer-built wrapper that emits the gen_ai.* attributes below.

Setup documented

Direct xAI Grok API calls (OpenAI-compatible API via openai SDK, no framework) instrumented by the telara-observability provider wrapper. Spans are tagged telara.span.kind=provider so the cost rollup excludes them when an active framework adapter already covers the same call (double-count guard). Uses the OpenAI SDK with base_url=https://api.x.ai/v1. pip extra: pip install telara-observability[openai]

Setup documented

Xero accounting platform for contacts, invoices, bills, accounts, payments, tracking categories, and financial review workflows.

Setup documented

Yammer enterprise social network

Setup documented

Zapier task usage remains admin CSV/dashboard — no programmatic per-member usage API confirmed for Connect (spend-gov research). Coming soon.

Setup documented

Zendesk customer service platform for managing tickets, users, and knowledge base articles

Setup documented

Zoom is a video communications platform that connects users through video, audio, chat, and webinars

Setup documented