TimescaleDB

Connect AI agents to TimescaleDB, the time-series database built on PostgreSQL. Agents run SQL, do CRUD without SQL, manage hypertables, run time_bucket rollups, and enforce chunk retention.

17 actions available

A stream of metrics or events arrives from an upstream service

Agent runs Insert Rows to batch-ingest the readings into a hypertable

Agent runs Time Bucket Query to roll the raw data into hourly averages

Agent reads the bucketed aggregates for the dashboard

Agent publishes the rollup to the dashboard sheet via Slack or Google Sheets

On schedule, agent runs Drop Chunks to enforce the retention window

Any Drop Chunks or Delete Rows that would remove data below the retention floor pauses for approval

What This Integration Enables

Agents ingest sensor, metric, and event streams into a hypertable one row or a batch at a time, roll raw data into hourly or daily summaries with `time_bucket` or continuous aggregates, enforce retention by dropping old chunks, and enable compression on historical data. Because TimescaleDB is wire-compatible with PostgreSQL, every standard SQL and row operation works alongside the time-series actions. Interval inputs such as `7 days` are validated to a safe character set before being placed into DDL and retention functions that cannot be parameterized.

Without FlowRunner

Raw data grows unbounded Time-series tables balloon with no retention or compression policy

Rollups computed by hand Someone writes a bespoke query each time a summary is needed

Deletes risk history A DELETE without a WHERE erases the whole hypertable

With FlowRunner

Retention as a flow step Drop Chunks enforces a retention window on schedule, with a count reported

Rollups on demand Time Bucket Query and continuous aggregates keep summaries fast

Guarded data loss Delete Rows needs a Where, and Drop Chunks needs an Older Than interval

Use Case Scenarios

Ingest, roll up, and publish

Metrics arrive from an upstream service. The agent calls Insert Rows to batch-ingest them into a hypertable, then Time Bucket Query to compute hourly averages, and publishes the rollup to a dashboard sheet. The summary that used to be a hand-written query runs as a step in the flow.

Compression once, retention on schedule

Early in a table's life the agent calls Enable Compression once to keep historical storage small. On a schedule, it calls Show Chunks to preview what is old, then Drop Chunks to drop chunks older than the retention interval, and reports how many chunks were dropped to the team channel.

Retention change with a human gate

Someone proposes tightening the retention window. Before the agent runs Drop Chunks with the new, shorter interval, it does not act on its own. It previews the affected chunks with Show Chunks, routes the preview for approval, and drops them only after a person confirms.

Human-in-Loop Highlight

Retention is the operation where a wrong interval quietly deletes history you meant to keep. FlowRunner's answer is human-in-the-loop, an execution pattern where the agent pauses on its own, assembles the context and the choices, routes to a human on their preferred channel, and resumes the moment they respond. When a Drop Chunks would remove chunks newer than the agreed retention floor, or a Delete Rows would affect more rows than the configured threshold, the agent pauses and asks through Slack: "Drop Chunks on `metrics` older than `30 days` would drop 61 chunks, the earliest from this quarter. Here is the Show Chunks preview. Approve, adjust the interval, or cancel?" The drop runs only after a person confirms, with the approver and timestamp captured in the run log. A connector can drop any chunk; an orchestration layer knows which drops should stop and ask.

Agent processes routinely

Detects exception requiring judgment

Clear match Continues automatically

Ambiguous Routes to human via Slack

Human decides

Agent resumes with decision

Agent Capabilities

17 actions

SQL

Execute Query Runs any SQL statement, including TimescaleDB functions, with `$1, $2` placeholders bound via the Parameters array, and returns rows, the affected count, and field metadata.

Rows

Select Rows Reads rows without writing SQL, with column selection, equality filters, ordering, and limit and offset.
Insert Row Inserts one row from a JSON object and returns the inserted row via `RETURNING *`.
Insert Rows Bulk-inserts an array of row objects in a single statement, which is ideal for batch ingest.
Update Rows Updates rows matching a non-empty Where object and returns the updated rows. A Where object is required.
Delete Rows Deletes rows matching a non-empty Where object and returns the deleted count. A Where object is required.
Upsert Row Runs `INSERT ... ON CONFLICT ... DO UPDATE` keyed by the given conflict columns.

Schema

Get Table Schema Returns column names, types, nullability, defaults, and positions from `information_schema.columns`.
List Tables Lists all tables and views in user schemas, with system schemas excluded.

Hypertables

Create Hypertable Converts an existing table into a time-partitioned hypertable, with an optional chunk time interval and optional migration of existing data.
List Hypertables Lists all hypertables with dimensions, chunk counts, and compression status.
Get Hypertable Chunks Lists the chunks of a hypertable with their time ranges and compression state.

Time-Series Analytics

Time Bucket Query Buckets rows into fixed time intervals and computes aggregates per bucket via `time_bucket`.

Compression & Retention

Enable Compression Turns on columnar compression, with an optional segment-by and order-by, and can schedule an automatic compression policy.
Create Continuous Aggregate Creates a `time_bucket` materialized view maintained incrementally, so summaries stay fast.
Show Chunks Lists a hypertable's chunks, optionally only those older than an interval, for previewing before a drop.
Drop Chunks Permanently drops chunks older than an interval for efficient retention. An Older Than interval is required.

Start building with TimescaleDB

$100 in credits. No card required. Connect in minutes.

Start Building Free Talk to us