# BotGauge Docs

> Learn how to red-team, trace, and evaluate AI agents with BotGauge. Setup guides, SDK and API reference, and guardrail examples for shipping safer agents.

- [BotGauge documentation](https://www.botgauge.com/docs/index.md): Find out how your AI agent fails before your users do. Red-team it, trace it, and manage your team with BotGauge. Start with a 5 to 15 minute quickstart.
- [Core concepts](https://www.botgauge.com/docs/get-started/core-concepts.md): The terms used across BotGauge docs: red-team runs, findings, held, sessions, turns, spans, projects, and workspaces, with one-line definitions.
- [Set up your account](https://www.botgauge.com/docs/get-started/setup.md): Sign in to BotGauge, finish workspace onboarding, understand where projects live, and choose whether to start with red teaming or tracing.
- [Red teaming overview](https://www.botgauge.com/docs/red-teaming/overview.md): How BotGauge red-teams your AI agent: what an attack is, what the results mean, the journey from connection to report, and what to prepare first.
- [Red teaming quickstart](https://www.botgauge.com/docs/red-teaming/quickstart.md): Explore a sample BotGauge red-team report in minutes, learn to read health, breaches, and evidence, then connect your own agent.
- [Connect your agent](https://www.botgauge.com/docs/red-teaming/connect-your-agent.md): Connect an AI agent to BotGauge as an API endpoint or with the BotGauge wrapper. Form fields, environments, secrets, cURL import, and the four connection checks.
- [Recon: let BotGauge learn your agent](https://www.botgauge.com/docs/red-teaming/recon.md): Recon is BotGauge's discovery chat that learns your agent's tools, data sources, and behavior. Covers depth, questions, context you can add, and reruns.
- [Rules and test plans](https://www.botgauge.com/docs/red-teaming/rules-and-plans.md): Review what BotGauge will test: the 20 standard rules, custom rules, plan settings such as run level and attempt budget, and framework-focused plans.
- [Approve, run, and watch the attack](https://www.botgauge.com/docs/red-teaming/approve-and-run.md): How BotGauge approvals work, why a start can be refused, what the prepared and adaptive attack rounds do, and how to watch or stop a run safely.
- [Agents and fleet health](https://www.botgauge.com/docs/red-teaming/agents.md): Read the BotGauge Agents page: fleet health cards, agent statuses, the agent table, and the Overview tab with health, open breaches, and Likely versus Confirmed.
- [Reports](https://www.botgauge.com/docs/red-teaming/reports.md): Every BotGauge red-team run creates a report. Read the Reports list, the result counts, rules, techniques, and attempts, and what the unlock hints mean.
- [Findings and evidence](https://www.botgauge.com/docs/red-teaming/findings.md): Read a BotGauge finding: what happened, the rule it breaks, the exact attack and reply, the tool call, the policy decision, and how it was decided.
- [Read your results](https://www.botgauge.com/docs/red-teaming/results.md): Understand BotGauge red-team results: how a breach is decided, the seven result types, severity, run verdicts, and how to read evidence for a finding.
- [Fix, verify, compare, and export](https://www.botgauge.com/docs/red-teaming/fix-and-verify.md): After a BotGauge red-team run: what to do with each result, verify a fix by replaying the attack, rerun errors, compare versions, and export reports.
- [Compliance view](https://www.botgauge.com/docs/red-teaming/compliance.md): See how BotGauge red-team results map to OWASP LLM Top 10, OWASP Agentic, MITRE ATLAS, and NIST frameworks. Item statuses, coverage, and key caveats.
- [Black box and grey box testing](https://www.botgauge.com/docs/red-teaming/black-box-and-grey-box.md): Black-box testing sees messages only. Grey-box adds tool calls, traces, and backend state through the BotGauge wrapper. Compare them and set up the wrapper.
- [Red teaming troubleshooting](https://www.botgauge.com/docs/red-teaming/troubleshooting.md): Fix common BotGauge red-teaming problems: failed connection tests, recon that hangs, refused starts, not-evaluable rules, incomplete runs, and low coverage.
- [Red teaming FAQ](https://www.botgauge.com/docs/red-teaming/faq.md): Answers to common questions about BotGauge red teaming: safety of your data, API keys, why rules are not evaluable, run time, and what a clean report means.
- [Red teaming glossary and quick reference](https://www.botgauge.com/docs/red-teaming/glossary.md): Definitions of BotGauge red-teaming terms such as recon, run pin, held, signal, canary, and verifier, plus the journey, results, and limits at a glance.
- [Tracing overview](https://www.botgauge.com/docs/tracing/overview.md): How BotGauge tracing works: standard OpenTelemetry over OTLP/HTTP, session and agent identity, privacy defaults, and the two ways to send traces.
- [Tracing quickstart](https://www.botgauge.com/docs/tracing/quickstart.md): Send your first AI agent trace to BotGauge in five minutes: create a project API key, install the Python SDK, and verify ingest with one command.
- [Get your API key](https://www.botgauge.com/docs/tracing/api-key.md): Create a BotGauge project API key, set BOTGAUGE_API_KEY, load it from a .env file, and avoid the 401 errors caused by using the wrong token.
- [Install the SDK](https://www.botgauge.com/docs/tracing/install.md): Install the BotGauge Python SDK with pip, add framework extras with botgauge setup, and pin versions in requirements.txt. Requires Python 3.10 or newer.
- [Manual tracing](https://www.botgauge.com/docs/tracing/manual-tracing.md): Use tracing.init(), tracing.turn(), and tracing.agent() to add BotGauge session and agent identity to every span, and learn how init() reads its settings.
- [Zero-code instrumentation](https://www.botgauge.com/docs/tracing/zero-code.md): Trace a Python AI agent without editing it. Launch your app with botgauge instrument, load a .env file, and gate CI on verified framework versions.
- [Capture and redaction](https://www.botgauge.com/docs/tracing/capture-and-redaction.md): Control what BotGauge exports from your agent. Choose metadata, text, or full capture, keep PII and credential masking on, and allow extra attribute keys.
- [Logs: view sessions, traces, and spans](https://www.botgauge.com/docs/observability/logs.md): Find the traces your app sends to BotGauge. Browse sessions, traces, and spans in Logs, read the columns, and open a session for messages and timeline.
- [Tracing troubleshooting](https://www.botgauge.com/docs/tracing/troubleshooting.md): Fix common BotGauge tracing problems: rejected API keys, nothing exported, missing session ids, skipped frameworks, dropped spans, and zero-cost traces.
- [Evals overview](https://www.botgauge.com/docs/evals/overview.md): Evals run an AI judge over your agent's traces and show pass or fail for every row in Logs. Learn judges, rules, results, and what each level reads.
- [Evals quickstart](https://www.botgauge.com/docs/evals/quickstart.md): Turn on your first judge and read your first result: open Judges, enable a judge, check Logs, and tune it. Includes three common judge setups.
- [Sample data](https://www.botgauge.com/docs/evals/sample-data.md): See eval results on day one: load BotGauge's sample traces, read the sample judge columns, and learn the limits on sample loads and your own judges.
- [Built-in judges](https://www.botgauge.com/docs/evals/built-in-judges.md): The 18 built-in BotGauge judges: what each checks, the levels it works at, and how it reports a result. Includes pass thresholds for score judges.
- [Custom judges](https://www.botgauge.com/docs/evals/custom-judges.md): Write a BotGauge judge for your own rule: choose scope and output type, use prompt variables, test it on real spans, and create it by API.
- [Rules and settings](https://www.botgauge.com/docs/evals/rules-and-settings.md): Control what BotGauge scores and when: sampling rate, wait before scoring, past data runs, filters, turning a judge off, and testing a judge on one example.
- [Reading results](https://www.botgauge.com/docs/evals/reading-results.md): What every cell in BotGauge Logs can say about an eval: scores, waiting, retrying, not sampled, not eligible, and how traces and sessions roll up spans.
- [Failures and troubleshooting](https://www.botgauge.com/docs/evals/failures-and-troubleshooting.md): How BotGauge retries a failed judge call, what each failure cause means, and fixes for common eval problems such as Settles, Not backfilled, and Queued.
- [API and limits](https://www.botgauge.com/docs/evals/api-and-limits.md): Run BotGauge evals from code: authentication, every eval endpoint, example requests for enabling, score-now, and results, plus monthly quota and size limits.
- [Supported frameworks](https://www.botgauge.com/docs/integrations/overview.md): BotGauge verifies tracing for 11 AI agent frameworks, including LangChain, OpenAI, Anthropic, CrewAI, Google ADK, and MCP. See extras and verified versions.
- [Trace LangChain with BotGauge](https://www.botgauge.com/docs/integrations/langchain.md): Trace LangChain agents with BotGauge, in code or with zero-code launch. Verified LangChain versions, install line, and a tested sample.
- [Trace LangGraph with BotGauge](https://www.botgauge.com/docs/integrations/langgraph.md): Trace LangGraph graphs with BotGauge, in code or zero-code. See verified versions, the install line, and a tested supervisor-graph sample.
- [Trace OpenAI SDK with BotGauge](https://www.botgauge.com/docs/integrations/openai.md): Trace OpenAI SDK calls with BotGauge in code or with zero-code launch. Verified versions, install line, and a tested chat completions sample.
- [Trace Anthropic SDK with BotGauge](https://www.botgauge.com/docs/integrations/anthropic.md): Trace Anthropic SDK calls with BotGauge in code or zero-code. Verified versions, install line, and a tested messages sample.
- [Trace LlamaIndex with BotGauge](https://www.botgauge.com/docs/integrations/llamaindex.md): Trace LlamaIndex apps with BotGauge in code or zero-code. Verified versions, install line, and a tested chat sample.
- [Trace CrewAI with BotGauge](https://www.botgauge.com/docs/integrations/crewai.md): Trace CrewAI crews with BotGauge in code or zero-code. Python 3.10 to 3.13, verified versions, and the OpenAI producer the extra adds.
- [Trace OpenAI Agents with BotGauge](https://www.botgauge.com/docs/integrations/openai-agents.md): Trace OpenAI Agents SDK runs with BotGauge in code or zero-code. Verified versions, install line, and a tested Runner sample.
- [Trace Google ADK with BotGauge](https://www.botgauge.com/docs/integrations/google-adk.md): Trace Google ADK agents with BotGauge in code or zero-code. Needs ADK 2.10 or newer, with a tested LiteLLM sample and known failure modes.
- [Trace AutoGen with BotGauge](https://www.botgauge.com/docs/integrations/autogen.md): Trace AutoGen AgentChat agents with BotGauge in code or zero-code. Verified version, install line, and a tested AssistantAgent sample.
- [Trace PydanticAI with BotGauge](https://www.botgauge.com/docs/integrations/pydantic-ai.md): Trace PydanticAI agents with BotGauge in code or zero-code. Python 3.11+ install line, the 3.10 fallback, and the token-count caveat.
- [Trace MCP with BotGauge](https://www.botgauge.com/docs/integrations/mcp.md): Trace MCP clients and servers with BotGauge in code or zero-code, including an LLM calling a tool on an in-process MCP server.
- [Versions not verified](https://www.botgauge.com/docs/integrations/versions-not-verified.md): What BotGauge does with framework versions it hasn't certified: traces them with a warning, skips known-broken ones, and lets CI enforce verified versions.
- [Any OTLP exporter](https://www.botgauge.com/docs/integrations/otlp.md): Send OpenTelemetry traces to BotGauge without the SDK. Endpoint, Bearer header, and gzip settings, plus Python, environment variable, and Collector setups.
- [Account and profile](https://www.botgauge.com/docs/platform/account-and-profile.md): Edit your BotGauge display name, manage email notifications for finished attacks and confirmed findings, and sign out of the console.
- [Organizations, workspaces, and projects](https://www.botgauge.com/docs/platform/concepts.md): How BotGauge organizes access: an organization contains workspaces, and each workspace contains projects with their own secrets and API keys.
- [Organization settings](https://www.botgauge.com/docs/platform/organization-settings.md): Rename or delete your BotGauge organization, review its workspaces and member counts, and understand who can change organization settings.
- [Roles and permissions](https://www.botgauge.com/docs/platform/roles-and-permissions.md): A complete matrix of BotGauge organization and workspace roles: who can invite people, manage projects and secrets, create API keys, and delete resources.
- [Members and invitations](https://www.botgauge.com/docs/platform/members-and-invitations.md): Invite people to a BotGauge organization or workspace, understand the 7-day invitation lifecycle, and change or remove access without losing history.
- [Workspaces and projects](https://www.botgauge.com/docs/platform/workspaces-and-projects.md): Create, rename, and delete BotGauge workspaces and projects, find their IDs, and understand Free plan limits and how to request more workspaces.
- [API keys](https://www.botgauge.com/docs/platform/api-keys.md): Create, list, and revoke BotGauge project API keys. Keys work with the SDK, the CLI, and the API, reach one project only, and are shown once.
- [Secrets](https://www.botgauge.com/docs/platform/secrets.md): Store named tokens for BotGauge agent connections. Reference secrets as $NAME in headers, replace values safely, and see what uses each secret before deleting.
- [CLI reference](https://www.botgauge.com/docs/reference/cli.md): Reference for the botgauge command: instrument, setup, doctor, probe, and help. Flags, exit codes, and the botg alias.
- [Environment variables](https://www.botgauge.com/docs/reference/environment-variables.md): Every BotGauge and OpenTelemetry environment variable the tracing SDK reads, with defaults, which command reads each, and how they interact.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.