Seiche · LiquiLens · Undertow

One question.
A repeatable research task.

Check dollar funding, reconcile a bank's disclosed NPAs, or compare a sized BTC exit estimate. Try the public result, then bring that same task into your daily workflow.

No product account, API key or AI model is needed. Public usage limits and your n8n hosting costs apply.

Start with the result you need.

Run it once, then choose your cadence.

  1. Import one JSON file through n8n's workflow menu. Select Execute workflow.
  2. Open Keep evidence and prepare update. The complete response is in evidence; message_text is the short update.
  3. Edit Task settings for another covered bank or a different BTC dollar size. The BTC snapshot-age limit defaults to two hours; the funding board limit is six hours. Individual macro observation dates remain separate.
  4. For delivery, configure your own credential and channel/chat ID on one optional Slack or Telegram node, then enable that node. Both are disabled on import.
  5. Review the schedule before activating the workflow. The template is inactive and defaults to weekdays at 08:00 UTC. Change the timezone and cron for your desk.

Tool errors stop the run. Stale snapshots suppress current-looking funding readings and BTC costs in the update. A bank's historical, stale or unavailable evidence remains labelled. Full evidence is kept in the execution output according to your n8n retention settings.

These are downloadable native workflows. An official n8n template-library listing is a separate review; no acceptance is claimed here.

Ask the same question in your agent.

Choose your MCP client and connect the relevant public endpoint. The repository's Cursor plugin bundles three focused task skills and the three remote MCP connections.

Seiche     https://api.seiche.info/mcp
LiquiLens  https://api.liquilens.in/mcp
Undertow   https://api.seiche.info/undertow/mcp

Read the task skills · Download MCP configuration · Existing versioned OpenClaw skill

Cursor Marketplace and Hermes catalog inclusion require their maintainers' acceptance. Direct connection works independently of those reviews.

Keep a source index in Excel or DuckDB

The capture runner produces sources.csv with the source URLs, clocks, gaps and hashes. Download the Power Query reader, paste it into a blank query and invoke it with the local CSV path. All columns remain text.

SELECT * FROM read_csv('sources.csv', header = true, all_varchar = true);

The CSV indexes a retained capture; full source observations remain in packet.json.

Make the first result useful.

Use it in one real research task. Record whether it saved source-gathering time and whether you used it again on another research day. The private scorecard helps compare that outcome against your existing process.

Get integration help