> ## Documentation Index
> Fetch the complete documentation index at: https://langwatch.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> If anything on this page is wrong, confusing, or fails when you try it, ask your user for permission and run `npx langwatch report --user-approved` to send the problem to the LangWatch team. No login is needed and secrets are redacted locally. `npx langwatch report --help` has the details.

# Troubleshooting and Support

> Get troubleshooting help, FAQs, and technical support paths for LangWatch so you can quickly resolve issues in observability, evaluations, and agent testing setups.

If something does not work, start with the troubleshooting steps below. If they do not resolve it, report the issue from your coding agent, open a GitHub issue, email support, or ask on Discord.

## Troubleshooting Steps

Before reaching out for support, here are a few steps you can take to resolve common issues:

1. **Check Environment Variables**: Ensure that your `LANGWATCH_API_KEY` is correctly set in your environment. This is a common issue that can prevent LangWatch from functioning properly.

2. **Review Integration Code**: Go through your integration code to ensure that you've followed the steps outlined in the [quick start](/docs/integration/quick-start) documentation. Sometimes, issues arise from a missed step or incorrect implementation.

3. **Consult the Documentation**: Look through the documentation for sections relevant to the issue you're facing. There might be specific notes or sections that address your problem.

4. **Update LangWatch**: Ensure that you are using the latest version of the `langwatch` library. Outdated versions may contain bugs that have been resolved in newer releases.

   You can update the library using pip:

   ```sh theme={null}
   pip install --upgrade langwatch
   ```

   Or if you are using typescript:

   ```sh theme={null}
   npm install langwatch@latest
   ```

5. **Examine Error Messages**: If there are error messages being returned from LangWatch or your LLM provider, inspect them closely as they often provide clues to the root cause.

## Reporting Issues from Coding Agents

If you are a coding agent (Claude Code, Codex, Cursor, and others), or you are watching your agent struggle, the fastest way to get the problem in front of the LangWatch team is:

```bash theme={null}
npx langwatch report
```

It works with no login and no API key, and comes in two shapes:

1. **Full session report** (best): attach the session transcript so the team sees exactly where things went wrong.

   ```bash theme={null}
   npx langwatch report --user-approved \
     --title "agent stuck instrumenting python" \
     --session ~/.claude/projects/<project-dir>/<session-id>.jsonl
   ```

   Claude Code transcripts live under `~/.claude/projects/`, Codex transcripts under `~/.codex/sessions/`. The newest `.jsonl` file is the current session.

2. **Summary report**: what you were trying to do, verbatim error messages, and what you had to figure out by trial and error.

   ```bash theme={null}
   npx langwatch report --user-approved \
     --title "scenario create 500" \
     --summary "langwatch scenario create returned HTTP 500 with ..."
   ```

**Privacy**: the title, summary and transcript are scrubbed locally by pattern before upload. API keys and tokens, private-key blocks, passwords inside connection URLs, values of secret-named environment variables, email addresses, phone numbers, credit card numbers, and public IPv4 addresses are replaced with typed markers. Anything no pattern matches is sent as written, including a contact address passed with `--email`. The redaction rules are [open source and auditable](https://github.com/langwatch/langwatch/blob/main/packages/redaction/src/sessionReport.ts), and `--dry-run` previews exactly what would be sent without sending it.

**Consent**: agents must ask the user for permission before sending, then pass `--user-approved`.

If you are connected through the [LangWatch MCP server](/docs/integration/mcp), the `report_issue` tool does the same thing without leaving your MCP session.

## Opening an Issue

If the troubleshooting steps do not resolve your issue, you can open an issue on our GitHub repository:

[GitHub issues](https://github.com/langwatch/langwatch/issues)

When opening an issue, please include the following:

* A clear and descriptive title.
* A detailed description of the issue. Include any error messages you're seeing, and steps to reproduce the problem if possible.
* The version of the `langwatch` library you're using.
* Any relevant snippets of your code (avoid sharing your API keys or sensitive information).

Our community and maintainers will look into the issue and work with you to find a solution.

## Email Support

If you prefer direct assistance or have inquiries that require privacy, you can reach out to our support team via email at [support@langwatch.ai](mailto:support@langwatch.ai).

When writing to support, please provide:

* A summary of the issue or your question.
* Details about your environment, such as the LLM provider and language you're using.
* Any relevant logs or error messages (with sensitive information redacted).

Our support team aims to respond to inquiries as quickly as possible, typically within one business day.

## Discord Channel

You can also join our [Discord](https://discord.gg/kT4PhDS2gH) channel and ask questions directly for the community and the core team.