Weaver.

A control plane for web-data pipelines. Detect a broken data contract, prepare a repair, and check the result before approving it.

01 / WORKFLOW STUDYDATA PIPELINES

A repair deserves
a second look.

THE REFERENCE POINT

What should
the data look like?

YAML contractMedian baseline

01Detect

Missing fields. Invalid values. Shifted data.

02Preview

Prepare a repair. Check affected fields.

03Approve

Review the proposal before applying a repair.

A bounded repair attempt, then another collection.

SHARED RECORD

Runs + incidents

SQLiteCLI / Dashboard / MCP

Weaver / System study

Contract checks, a repair preview, and a shared record of what happened.

Original diagram · based on public source

01 / THE CONTRIBUTION

My part in the project.

Project author. The README credits Kiro and Claude assistance; automated data commits are also recorded in the repository.

02 / INSIDE THE BUILD

Combine YAML contracts with median baselines to check for missing fields, invalid values, and shifted data.
Check repaired fields in a preview before approval, then recollect data after a bounded repair attempt.
Store runs and incidents in SQLite for the CLI, web dashboard, and MCP tools to share.

03 / STATUS & LIMITS

The default experience replays fixtures. Live collection and repair require Bright Data configuration.
A public dashboard deployment and production reliability measurements are not documented in the reviewed sources.

04 / KEEP EXPLORING

Explore the code and follow the latest state of the project in its repository.

About these project notes

Based on the public README, drift engine, preview checks, and repair workflow as of September 2026.