Engineering teams tend to let internal documentation rot because nobody opens a page until something breaks. Oxynote, a new open-source wiki released under Apache-2.0, tries to fix that incentive problem by letting a page carry live data next to its explanatory text. A page can embed a chart that runs its own query against Prometheus, PostgreSQL, MySQL or MariaDB — so a runbook showing replication lag or slow query counts sits directly beside the steps for handling them.
For PostgreSQL users specifically, Live Metrics blocks execute SQL against your databases and render the result as a time series, bar chart or gauge. The details that matter:
- Queries support Grafana-style macros such as
$__timeFilter,$__timeGroupand$__timeFrom, so existing dashboard queries can be carried over largely as-is. - Oxynote connects only to read-only data sources: a role or database with
default_transaction_read_onlyenabled, or a standby server. - Every session additionally runs in read-only mode as a second layer of protection.
- Oxynote stores its own data in PostgreSQL 18, which is embedded in the all-in-one Docker image.
- Edits to important pages follow a pull-request-like flow: a draft, then a diff, then approval — meaning a chart and its surrounding explanation are reviewed together.
The design rationale is straightforward: documentation goes stale because people open it only when something breaks. Live data gives the team a reason to open a page every day, and pages read daily stay up to date. For developers, the practical appeal is that query logic and operational instructions live in one reviewable artifact rather than drifting apart between a dashboard and a wiki. The read-only constraints also mean the wiki can be pointed at production replicas without granting it write access, which keeps the risk of an embedded query low.