TraceKit DocsAI Setup
Set up TraceKit with AI assistants like Claude Code, Cursor, and Codex. Install shared skills, enable MCP, and get guided setup for any SDK.
AI assistants can guide you through full SDK setup, configuration, and feature enablement. Install the shared TraceKit skills, then enable MCP if your assistant supports live TraceKit tools.
Install AI Skills
TraceKit provides AI skills that teach your assistant how to integrate every SDK, configure features, and debug issues. Choose your tool below to install.
Claude Code
From inside the cloned repo, install the TraceKit plugin and shared skills in Claude Code.
/install-plugin https://github.com/tracekit-dev/tracekit-for-aiCursor
From inside the cloned repo, use this repository as the Cursor plugin source. Add it in Cursor settings as a plugin or doc source, depending on your Cursor version.
tracekit-dev/tracekit-for-aiCodex
From inside the cloned repo, load the local marketplace file and install the TraceKit plugin. Codex reads the shared skills and bundled MCP wiring through the TraceKit plugin.
.agents/plugins/marketplace.jsonnpx skills add
Universal install command that works with any compatible AI assistant. Great for shared skills. MCP depends on whether the tool also supports loading project MCP config.
npx skills add tracekit-dev/tracekit-for-aiInstall MCP
TraceKit also ships an MCP server so supported assistants can query live services, traces, dashboards, alerts, and triage data. The first run automatically downloads the correct tracekit-agent binary from the latest GitHub release, and it checks for updates hourly so you always have the latest version.
Prerequisite
Clone the TraceKit AI repo first. The MCP commands below use the absolute path to the launcher script, so you can run them from anywhere.
git clone https://github.com/tracekit-dev/tracekit-for-ai.gitReplace /path/to with the absolute path where you cloned the repo.
Claude Code
--scope project scopes the MCP server to the current workspace. Use --scope user to make it available everywhere.
claude mcp add --scope project tracekit /path/to/tracekit-for-ai/scripts/run-tracekit-mcp.shCursor
Add a stdio MCP server in Cursor settings pointing to the launcher script.
/path/to/tracekit-for-ai/scripts/run-tracekit-mcp.shCodex
Add TraceKit as a global MCP server in Codex.
codex mcp add tracekit -- /path/to/tracekit-for-ai/scripts/run-tracekit-mcp.shAvailable MCP tools (19)
Observability: tracekit_status, tracekit_dashboard, tracekit_services, tracekit_service_detail, tracekit_service_errors, tracekit_traces, tracekit_trace_detail, tracekit_releases.
Alerts: tracekit_alert_rules, tracekit_alert_history, tracekit_create_alert_rule, tracekit_delete_alert_rule, tracekit_toggle_alert_rule, tracekit_channels.
Triage: tracekit_triage_inbox, tracekit_triage_acknowledge, tracekit_triage_investigate, tracekit_triage_resolve, tracekit_triage_snooze.
Authentication
The MCP server reads credentials from ~/.tracekitconfig or the TRACEKIT_API_KEY environment variable. You can authenticate with the built-in email verification flow below, or set the env var directly if you already have an API key.
Start Auth
/path/to/tracekit-for-ai/scripts/run-tracekit-auth.sh register --email you@example.comVerify Code
/path/to/tracekit-for-ai/scripts/run-tracekit-auth.sh verify --session-id <session_id> --code <code>If the email already has a TraceKit account, verification signs you in. Otherwise, a new account is created automatically.
Backend SDKs
Copy a prompt below and paste it into your AI assistant to set up a backend SDK.
| SDK | Prompt |
|---|---|
| Go | Set up TraceKit tracing in my Go project |
| Node.js | Set up TraceKit tracing in my Node.js project |
| Python | Set up TraceKit tracing in my Python project |
| PHP | Set up TraceKit tracing in my PHP project |
| Laravel | Set up TraceKit tracing in my Laravel project |
| Java | Set up TraceKit tracing in my Java project |
| .NET | Set up TraceKit tracing in my .NET project |
| Ruby | Set up TraceKit tracing in my Ruby project |
Frontend SDKs
Copy a prompt below and paste it into your AI assistant to add frontend error tracking and tracing.
| SDK | Prompt |
|---|---|
| Browser | Add TraceKit error tracking to my web app |
| React | Set up TraceKit tracing in my React app |
| Vue | Set up TraceKit tracing in my Vue app |
| Angular | Set up TraceKit tracing in my Angular app |
| Next.js | Set up TraceKit tracing in my Next.js app |
| Nuxt | Set up TraceKit tracing in my Nuxt app |
Features
Enable advanced TraceKit features with a single prompt.
| Feature | Prompt |
|---|---|
| Session Replay | Enable session replay in my TraceKit setup |
| Source Maps | Set up source map uploads for my TraceKit project |
| Release Tracking | Configure release tracking with TraceKit |
| Alerts | Set up alerts for my TraceKit project |
| Distributed Tracing | Enable distributed tracing between my frontend and backend |
| Code Monitoring | Set up code monitoring breakpoints with TraceKit |
llms.txt
TraceKit provides machine-readable content indexes for AI assistants and LLMs.
/llms.txt-- Concise index of all TraceKit documentation and skill URLs. Lightweight for quick context loading./llms-full.txt-- Expanded version with full skill content inlined. Ideal for Cursor @docs.new ingestion.