The Copilot streams tool-calling responses against the same data the dashboard shows — events, delivery, quality, insights, audiences, billing. The MCP server exposes those tools to your editor, so an agent can investigate without context-switching.
Meta reports a soft drop in attributed revenue. You open the dashboard. Then the events stream. Then the connector logs. Then GTM. Twenty minutes in, you still haven't formed a theory — you've only been collecting tabs.
Marketing wants confirmation that the new bundle SKU is firing purchases with line items, value, currency, and EMQ above 7. The SQL exists somewhere. The screenshot is older than the campaign. The launch slips by a day.
"Are we losing match quality on iOS?" You build the answer in three queries, two pivots, and a Loom. Repeat next week. The analyst-hour gap between asked and answered is where most operational debt lives.
Make the pipeline answer in the same surface where the question gets asked — chat, editor, terminal. Twelve tools, one MCP server, live data. The Copilot does the same job your senior analyst does when they're sat next to you.
Pipeline KPIs across events, delivery, latency, and revenue.
Per-platform delivery rate and volume by reporting window.
Match quality, identifier coverage, and validation issue trend.
Revenue and conversions by campaign, source, or medium.
Granted, denied, partial, and GPC consent counts.
Full event payload, delivery attempts, retries, and hashed PII keys.
Customer-level scoring from the active merchant model.
Audience members with hashed identifiers for spot checks.
Audience definitions, sync targets, and freshness.
Natural language to audience SQL with size estimate.
Live event stream filtered by platform, status, and event name.
Synthetic event through the delivery pipeline.
Health, latest event, p95 latency, and 24h error rate.
Search ingestion and delivery errors by severity and query.
Queue S3, BigQuery, or Snowflake exports.
Active detector findings across drops, drift, and regressions.
Seasonal baseline investigation for anomaly root cause.
Suggested field mapping for schema drift.
Acknowledge a finding with a durable reason.
Deletion DSAR status by email, phone, or external id.
Plan usage, included events, and soft overage state.
Plan cap, renewal timing, and AI Copilot budget.
Run anomaly detectors on demand before go-live.
Create recurring report email schedules.
$ npx @tracklayer/mcp-server --api-key tl_<your_workspace_key> ▸ resolved 24 tools ▸ verified workspace · glass-house.co ▸ listening on stdio · ready for client
Add to settings.json under mcpServers. Authenticate once, query in any project session.
Settings → MCP → Add. Tools appear inline in chat — Cursor calls them automatically.
claude_desktop_config-style JSON. Tools surface as actions during conversations.
Add a mcpServers entry to ~/.continue/config.yaml. Tools resolve in the chat panel.
Add to cline_mcp_settings.json. Read-only by default; opt into write tools per workspace.
Wire @tracklayer/mcp-server into the Replit shell alongside your agent. Same npx command.
We use essential cookies to keep the site secure and functional. Analytics and third-party tags run only with your consent. See our Cookie Policy.
We use essential cookies to keep the site secure and functional. Analytics and third-party tags run only with your consent. See our Cookie Policy.
