Claude Code
$ claude mcp add --transport http anectico https://app.anectico.com/mcpThen sign in once in the browser. No key to copy.
Observability and product analytics your coding agent can operate.
Every signal about your product, tied to the person behind it. Ask Claude Code or Codex why.
Connected in one command. Investigating in the next.
No dashboard tour, no query language to learn. Your agent discovers the operations, reads the same evidence a person would, and cites it back to you.
Codex
$ codex mcp add anectico --url https://app.anectico.com/mcpThen sign in once in the browser. No key to copy.
Anectico CLI
$ anectico login
$ anectico persons timeline <person-id>Every operation the agent has, from your terminal or CI.
Then ask, in your own words:
- “Set up observability for this project with Anectico.”
- “Which customers hit errors in the last hour, and what did they do just before?”
- “Measure signup → first deploy within 24 hours for the last week and show me who stalled.”
- “Is anything firing? Acknowledge it and open an incident if the trace confirms it.”
One identity spine under everything
Most tools keep errors in one product, analytics in another and replay in a third, and leave you to guess whether they are the same person. Anectico resolves identity once, at ingest, and hangs every signal on it. A funnel step opens the people in it. A person opens their trace. A trace opens the replay of the session that made it.
The whole picture, not a bundle of tabs
Issues with the people behind them
Errors group into issues ranked by recurrence and customer impact, with the affected people and accounts on the row, not in a report.
Traces, logs and metrics
Follow one request across services, read its logs beside it, and search structured fields with live tail when you need it now.
Session replay with the evidence alongside
Watch what the person saw, with their console errors, network calls and the exact event they fired next.
Product analytics that draw their caveats
Trends, ordered funnels and exact-period retention, written as sentences. Coverage, maturity and identity resolution travel with every number.
Alerts, anomalies and incidents
Rules on any measure, anomaly findings you can act on, incidents with on-call and escalation, all readable and actionable by an agent.
Your agents on the map too
A fleet inventory of what is running, run-level quality, two-person containment and a content policy for what recorded data may be used for.
Illustrative previews.
Honest by design
Every result carries its coverage, its maturity and its identity resolution. A dashboard cannot make a claim the data does not support, and neither can an agent reading it.
- A zero is never “no coverage”.
- Unseen is reported as unknown, never as nothing.
- Pending is never drop-off.
- Only closed, fully observed windows count as mature.
- Withheld is never “none recorded”.
- Missing permission is stated as such, not as an absence.
- Stopped publishes nothing.
- No partial result ever looks like a finished one.
Get a seat in the private preview
We are onboarding a small number of teams while the analytics engine finishes its capacity qualification. Leave a work email and we will write when a seat opens.
Have a specific project in mind? Request access for a team.