# Kensa

> Kensa detects agent issues in production traces, gives coding agents evidence to investigate and fix them, and monitors later production traffic for recurrence after a fix is deployed.

This site serves the hosted Kensa product. Its integrations use the `kensa-sdk` package and a project-scoped MCP connection.
Resolve the relative links below against the origin serving this file. Use that same origin for the web application, setup instructions, and MCP endpoint. The trace-ingestion endpoint can use a different origin; obtain it from `get_setup_status` rather than inferring it from the website URL.

## Start here

- [Agent setup instructions](/install): The setup skill for connecting a coding agent, configuring tracing, and verifying that a specific application trace arrived.
- [Sign in](/login): Access projects, issues, trace evidence, and fix monitoring.
- [Home](/home): Open an accessible project or the project picker; requires sign-in.
- [Pricing](/pricing): Current plans and usage information.

## Docs

Each page is also served as markdown at the link below. [All docs in one file](/llms-full.txt) joins every page.

- [Introduction](/docs/introduction.md): Kensa finds and fixes issues in your agents.
- [Quickstart](/docs/quickstart.md): Connect Kensa, check the connection, and explore your issues.
- [Concepts](/docs/concepts.md): The few terms you need to use Kensa.
- [Issues](/docs/issues.md): Look at issues and know what to do next.
- [Agent workflow](/docs/agent-workflow.md): Give your coding agent an issue, deploy the change, and report the fix.
- [Connect your coding agent](/docs/mcp/connect.md): Give your coding agent access to Kensa through MCP.
- [Tool reference](/docs/mcp/tools.md): Inputs and behavior for Kensa's MCP tools.
- [Python SDK](/docs/sdk/python.md): Send traces to Kensa from a Python app.
- [TypeScript SDK](/docs/sdk/typescript.md): Send traces to Kensa from a Node.js app.
- [Tracing](/docs/tracing.md): Send traces to Kensa from Langfuse, the Kensa SDK, or OpenTelemetry.
- [Privacy](/docs/privacy.md): How Kensa handles your keys and trace data.
- [Changelog](/docs/changelog.md): What's new in Kensa.

## Production workflow

1. Connect an existing Langfuse source or export OpenTelemetry traces over OTLP/HTTP protobuf. The optional `kensa-sdk` packages on npm and PyPI help configure tracing for TypeScript/Node and Python applications.
2. Kensa normalizes and redacts trace data before persistence, then detects issues and groups repeated occurrences with supporting trace and span evidence.
3. Review issues in the web application or let a coding agent read them through MCP. The coding agent investigates and changes the application's repository using its own tools.
4. After a fix is deployed, report its deployed revision and environment through MCP. Kensa checks later production traces for recurrence; reporting a fix alone does not prove it worked.

## Coding-agent integration

- [MCP endpoint](/api/mcp): HTTP MCP with OAuth authorization scoped to one project. Connect through the coding client's native MCP support and follow the browser consent flow.
- [Agent connections](/oauth/connections): Review or revoke coding-agent access; requires sign-in.

Available MCP tools:

- `get_setup_status`: Read connection readiness and the project's OTLP endpoint. Supply the original trace ID to verify that a specific application trace was received.
- `list_issues`: List detected issues for the authorized project.
- `get_issue`: Read an issue and its supporting evidence for investigation.
- `report_fix`: Report an already deployed fix so subsequent production traffic can be checked for recurrence.

MCP OAuth credentials and trace-ingestion API keys serve different purposes. Use the project's tracing key for ingestion, never as an MCP OAuth substitute. Keep credentials in the application's secret store, outside source control and chat.

For fix monitoring, traces must identify the deployed revision with `service.version` and the environment with `deployment.environment.name`. Use actual deployment metadata. A local trace can verify ingestion without establishing production-monitoring readiness.

## Optional

- [Security](/security): Trace redaction, scoped access, credentials, retention, and vulnerability reporting.
- [Privacy](/privacy): Data handling and privacy terms.
- [Terms](/terms): Terms for using the hosted service.
- [Support](mailto:support@kensa.sh): Product and integration questions.