Oden

What you can build

• Name the reason for each downtime

  States in the downtime category without a reason are listed each shift. The operator drafts a reason from nearby metrics and events and proposes /v2/interval/update, one state at a time.

  With Line Downtime

• Watch OEE by line, daily

  An OQL query groups aggregation.oee by line.name with bucket 1D. The operator flags lines whose trend falls and attaches the availability, performance, and quality components it read.

  With OEE Loss, KPI Drift

• Raise a work order from a repeating fault

  When the same state reason recurs on a line, the operator proposes a maintenance work order with an external_id for your CMMS and the interval ids it used.

  With Predictive Maintenance

• Catch scrap outside its target

  Scrap per batch is compared with the run's target and its thresholds. Above the upper limit, the operator proposes a review note and files the scrap records behind it.

  With Scrap Rate, Batch Deviation

Requirements

• An Oden organization and an API token that Oden issues for it
• Line and factory ids or names; some endpoints must be called per line or factory, so request counts grow with the number of lines
• Timestamps in ISO 8601 form, YYYY-MM-DDTHH:MM:SSZ, on every POST body

Authentication

A token issued by Oden, sent on every request as Authorization: Token <credentials>; an invalid token returns 401 and one without access to the data returns 403.

Limits

• Oden states it does not rate limit today but will, and asks clients to stay at or under one request per second
• The v2 specification lists no webhook or push endpoint; intervals and states are found by polling with start_time and end_time
• Scrap and yield for an interval is the sum of its records; correct it with an offset record or by updating one
• OQL does not support plain UNION, or GROUP BY with WINDOW in one query; use UNION ALL or a helper query

Questions and answers

What does an Oden request look like?

A POST to https://api.oden.app/v2/{object}/{action}, for example /v2/factory/search, with a JSON body and Authorization: Token <credentials>. POST is the only HTTP method: search reads, set creates or updates, delete removes. Responses are JSON with ISO 8601 timestamps and durations.

What is OQL?

Oden Query Language, a SQL-like language for factory data. Every query starts with aggregate, takes a where time from ... to ... range, and can group by line.name or bucket 1h. It runs through POST /v2/oql/query and in the OQL Tester and Forge inside the Oden platform.

Does Oden rate limit?

Not under the current implementation, according to the API description. Oden says rate limiting will be introduced and recommends clients stay at or below one request per second. The OQL query endpoint already defines a 429 Too Many Requests response.