Add analytics to Continue.
Continue reads MCP servers from ~/.continue/config.yaml, or from a block file under .continue/mcpServers/, where mcpServers is a YAML list and not an object. Add smolanalytics with your org token and Continue can answer funnels, retention, paths and what broke today from your own events, in VS Code or JetBrains, and open the instrumentation PR that produced those events in the first place.
mcpServers:
- name: smolanalytics
type: streamable-http
url: https://smolanalytics.com/api/mcp
apiKey: YOUR_ORG_TOKEN
# apiKey is sent as "Authorization: Bearer YOUR_ORG_TOKEN".
# Longhand, if you would rather set the header yourself:
# requestOptions:
# headers:
# Authorization: Bearer YOUR_ORG_TOKENmcpServers is a list in Continue, so the leading dash matters. apiKey and requestOptions exist only on the sse and streamable-http shapes, not the stdio one. The same block works as .continue/mcpServers/smolanalytics.yaml if you keep blocks in their own files (those also need name, version and schema: v1 at the top). Self-hosting? Swap url and apiKey for command: smolanalytics, args: [mcp].
how do I add analytics to continue?
Continue reads MCP servers from ~/.continue/config.yaml, or from a block file under .continue/mcpServers/, where mcpServers is a YAML list and not an object. Add smolanalytics with your org token and Continue can answer funnels, retention, paths and what broke today from your own events, in VS Code or JetBrains, and open the instrumentation PR that produced those events in the first place.
Continue is the open-source assistant, and its config is a file you can commit, which makes the MCP entry reviewable like any other change: the team gets the same analytics connection from the same commit instead of five people wiring five configs by hand. It also runs in JetBrains, so this is how you ask your product numbers from IntelliJ, GoLand or PyCharm, where most analytics tooling assumes you live in VS Code. MCP tools surface in Agent mode, so ask there: "retention for users who hit the pricing page, last 30 days".
The header is the part worth knowing. Continue's reference page lists requestOptions for sse and streamable-http servers without spelling out what goes inside it, so people reach for an mcp-remote stdio bridge they do not need. Two things work natively: apiKey is sent as an Authorization Bearer header, and requestOptions.headers sets any header outright. One org token covers everything: pass project="acme" to any of the 79 tools and it routes to that project's instance, while the read key stays server-side, so the only secret in your config is the org token.
Instrumenting comes first. Connect your GitHub repo and we open an instrumentation PR you review and merge, or add the script tag yourself. After that, every number Continue reads is a deterministic report that a CI test pins to the dashboard and the HTTP API, and the answers come from whichever model you already pay for in Continue, so nobody meters your AI. Open source (MIT), self-host free or $49/mo for 1M events.
Honest pricing: 14-day full trial, no credit card. Then Pro $49/mo, never metered on seats or sites. Overage is $8/million with an emailed receipt, the dashboard never locks, and self-hosting the single Go binary is free forever (MIT).
Add analytics to Continue tonight.
One snippet, or one endpoint. Tomorrow morning the verdict tells you which part of the funnel to fix.