# Moda

> Moda is a harness engineering platform: it turns production traces into verified improvements for your agent harness (prompts, tools, skills, evals, memory). Send traces — the full record of each agent run — through the Moda SDKs, OpenTelemetry, or the HTTP API; Moda analyzes them for emotion, tool-failure, laziness, and hallucination signals, groups them into use cases, and surfaces recurring problems. Everything in the dashboard is also available through the Data API and CLI.

## Documentation — Get started

- [Introduction](https://docs.moda.dev): Moda turns production traces into verified improvements for your agent harness: prompts, tools, skills, evals, memory. Ingest traces, explore the analysis in a dashboard, or query it over an API.
- [Quickstart](https://docs.moda.dev/quickstart): Install the Moda SDK, capture your first OpenAI or Anthropic call, and see the trace in the dashboard.

## Documentation — Concepts

- [Data model](https://docs.moda.dev/concepts/data-model): The objects Moda stores — traces, messages, users, environments — and the fields that control grouping and prompt attribution.
- [Analysis](https://docs.moda.dev/concepts/analysis): What Moda computes from ingested traces and where each result appears in the dashboard and Data API.

## Documentation — Dashboard

- [Analytics](https://docs.moda.dev/dashboard/overview): The Analytics page summarizes one agent's health, activity, quality rates, and what needs review over a selected time window.
- [Traces](https://docs.moda.dev/dashboard/traces): Browse every ingested trace, and inspect a single agent run as a message timeline or as its extracted world state.
- [Signals](https://docs.moda.dev/dashboard/signals): Per-trace detections — user emotion, tool failures, laziness, and hallucinations — with KPIs, taxonomies, and evidence links.
- [Problems](https://docs.moda.dev/dashboard/problems): Root-cause groups of negative signals with an Open/Reopened/Fixed/Resolved lifecycle and a built-in feedback loop.
- [Use Cases](https://docs.moda.dev/dashboard/use-cases): A hierarchical taxonomy of what users ask your agent, kept current automatically and browsable down to individual traces.
- [Users and Tool Catalog](https://docs.moda.dev/dashboard/users-and-tools): Per-user health and activity history, and an inventory of every tool your agent calls with effect classification and failure stats.

## Documentation — Administration

- [Authentication](https://docs.moda.dev/administration/authentication): Create API keys, use the right auth header for each API, and authenticate the CLI.
- [Members and roles](https://docs.moda.dev/administration/team): Invite teammates to your workspace and control what each role can do.
- [Limits](https://docs.moda.dev/administration/limits): Request size, batch, and query limits across the Ingestion API, Data API, dashboard, and CLI.

## Ingestion — Send data

- [Ingestion overview](https://docs.moda.dev/ingestion/overview): Choose how to send LLM and agent telemetry from your application to Moda.

## Ingestion — SDKs

- [Moda SDK for Python](https://docs.moda.dev/ingestion/python): Install moda-ai to automatically capture OpenAI and Anthropic calls with trace and user context.
- [Moda SDK for Node.js](https://docs.moda.dev/ingestion/node): Install moda-ai to automatically capture openai and @anthropic-ai/sdk calls with trace and user context.

## Ingestion — Providers

- [OpenAI](https://docs.moda.dev/ingestion/providers/openai): Capture OpenAI API calls automatically with the Moda SDK for Python or Node.js.
- [Anthropic](https://docs.moda.dev/ingestion/providers/anthropic): Capture Anthropic Messages API calls automatically with the Moda SDK for Python or Node.js.
- [OpenRouter](https://docs.moda.dev/ingestion/providers/openrouter): Capture OpenRouter calls made through the OpenAI client with the Moda SDK for Python or Node.js.
- [Azure OpenAI](https://docs.moda.dev/ingestion/providers/azure-openai): Capture Azure OpenAI calls automatically with the Moda SDK for Python; from Node.js, use OpenTelemetry, the HTTP API, or manual tracing.
- [Amazon Bedrock](https://docs.moda.dev/ingestion/providers/bedrock): Capture Anthropic models on Bedrock automatically with the Python SDK, or send other Bedrock traffic to Moda through manual capture or the HTTP API.
- [Vapi](https://docs.moda.dev/ingestion/providers/vapi): Forward Vapi end-of-call report webhooks to Moda to capture voice calls with their turns, tool calls, and transfers.
- [OpenClaw](https://docs.moda.dev/ingestion/providers/openclaw): Export OpenClaw's OTLP telemetry to Moda and generate the required configuration with the Node.js SDK helpers.

## Ingestion — Frameworks and agents

- [Vercel AI SDK](https://docs.moda.dev/ingestion/vercel-ai-sdk): Capture generateText, streamText, generateObject, and streamObject calls by passing Moda's telemetry configuration to the AI SDK's experimental_telemetry option.
- [Claude Agent SDK](https://docs.moda.dev/ingestion/claude-agent-sdk): Install moda-claude-agent-sdk to capture Claude Agent SDK runs with prompts, responses, token usage, and tool-call counts.
- [Coding agents](https://docs.moda.dev/ingestion/coding-agents): Send Claude Code, Codex, and Cursor session telemetry to Moda's coding-agent ingestion endpoints.

## Ingestion — OpenTelemetry

- [OpenTelemetry](https://docs.moda.dev/ingestion/opentelemetry): Point an existing OpenTelemetry exporter or collector at Moda's OTLP trace endpoint and control attribution with span attributes.

## Ingestion — API reference

- [Ingestion API](https://docs.moda.dev/ingestion/http-api): HTTP reference for the Moda Ingestion API: authentication, POST /v1/ingest, POST /v1/ingest/multi, and POST /v1/traces.
- [Reliability](https://docs.moda.dev/ingestion/reliability): Error envelope, status codes, retry guidance, and limits for the Moda Ingestion API.

## Data API & CLI — Data API

- [Data API overview](https://docs.moda.dev/data-api/overview): Read traces, signals, clusters, and problems over HTTP with an API key — the same data the dashboard shows.
- [Search](https://docs.moda.dev/data-api/search): GET /search — keyword, semantic, and hybrid search over individual messages, with anchors for jumping to trace context.
- [Traces](https://docs.moda.dev/data-api/traces): GET /conversations for filtered trace lists, /context for windowed messages, /step-scores for outcome scores, and /world-state for slots, threads, and state events.
- [Signals](https://docs.moda.dev/data-api/signals): GET /frustrations, /emotions, /tool-failures, and /hallucinations — signal detections with evidence and embedded trace context.
- [Analytics](https://docs.moda.dev/data-api/analytics): GET /overview for headline KPIs, /clusters for the use-case hierarchy, and /task-clusters/search for finding cluster nodes.
- [Problems](https://docs.moda.dev/data-api/problems): GET /problems for the ranked root-cause list, per-problem detail and evidence, and the feedback endpoints that steer discovery.

## Data API & CLI — CLI

- [CLI overview](https://docs.moda.dev/cli/overview): Install the Moda CLI, authenticate, provision API keys headlessly, and understand profiles, output formats, and exit codes.
- [CLI reference](https://docs.moda.dev/cli/reference): Moda setup, data and investigation commands with flags and example output.
- [Using the CLI from agents and CI](https://docs.moda.dev/cli/agents): Drive the moda CLI from coding agents and CI pipelines: agent envelopes, exit-code handling, headless provisioning, and runnable examples.

## Data API & CLI — MCP server

- [MCP server](https://docs.moda.dev/mcp/overview): Connect any MCP client to Moda's hosted MCP server, which exposes every remote-capable moda CLI command as a tool authenticated per request with your API key.
- [MCP tool reference](https://docs.moda.dev/mcp/tools): Moda data, investigation, and connection tools with parameters and examples.
- [CLI parity map](https://docs.moda.dev/mcp/parity): How the documented MCP tools map onto moda CLI commands, where wire behavior is identical, where it deliberately deviates, and which commands stay CLI-only.