Post notifications from dbt, Terraform, Argo CD, AWS, or any flow task. Schedule reports into channels as editable Canvases. Gate destructive steps with Block Kit approvals. Spin incident channels with on-call invited. Export channel history to your data warehouse.
Slack is where teams already work. Kestra makes every other tool reach them there: notifications from dbt, reports from Snowflake, approval gates for Terraform, incident channels spun from any failure, message archives exported to BigQuery. Slack stops being a per-tool integration target and becomes the meeting point for the whole stack.
Every flow can route its completion or failure into Slack as the last step. dbt build finished, app.chats.Post fires. Terraform applied, app.chats.Post fires. AWS Lambda errored, app.chats.Post fires. Argo CD reported Healthy, app.chats.Post fires. The Slack notification task is the same regardless of the upstream tool. No per-tool integration code, no per-tool webhook config, no per-tool YAML.
A scheduled flow queries any data source (Snowflake, BigQuery, AWS Athena, NetBox, GitHub stats, AWS Cost Explorer), formats the rows into markdown sections, then app.canvases.Create posts an editable Canvas to the right channel. Leadership reads the daily metrics during standup, adds context inline, and the next morning's run updates the Canvas. Reports stop being PDFs nobody opens.
Before any destructive step in any flow (terraform apply, kubectl Apply, Argo CD sync, Snowflake DELETE, IAM key rotation), app.chats.Post sends a Block Kit message with the change summary and Approve / Reject buttons. The Pause task waits. The click hits the Slack Events trigger; a handler flow resumes the original execution with the reviewer identity recorded. One gate pattern, every cross-tool chain.
When a production flow fails (dbt, Terraform, Argo CD, Kubernetes, AWS, NetBox, anything), the errors branch fires app.conversations.Create with a dated channel name, app.conversations.Invite adds on-call from PagerDuty or NetBox lookup, app.canvases.Create posts the runbook for this failure type, and app.chats.Post kicks off the channel. Same pattern across every monitored flow.
app.core.Trigger receives slash commands like /deploy, /rollback, /restart from Slack's Events API. Same trigger captures interactive component clicks (Block Kit buttons). Slack joins webhooks, schedules, S3 triggers, Kafka triggers, and Git events as one valid way to launch a flow. The full event-driven catalog stays open.
app.conversations.History pulls channel messages into Kestra storage. From there, the BigQuery or Snowflake plugin lands the result in your warehouse for support-channel analytics, response-time SLAs, top-question clustering. app.users.List enumerates the workspace for identity reconciliation with HubSpot, Okta, or NetBox contacts. app.files.Info inventories shared files for compliance reviews.
Patterns engineering, data, and ops teams ship today. Each one shows Slack connecting two systems beyond itself, with the real plugin classes in play.
A reusable notify subflow takes a channel ID, a message template, and a status as inputs. Parent flows (dbt build, terraform apply, argocd sync, aws lambda invoke) call the subflow on success or failure. Per-tool channel routing (#data-pipelines for dbt, #infra for Terraform, #releases for Argo CD, #aws-alerts for Lambda) lives in the parent. The notification logic and the Block Kit formatting live once, in the subflow.
Adding a new upstream tool means one line in the parent flow. The notify subflow does not change.
No Slack admin maintenance when a new team wants its own channel. The parent flow picks the channel.
Status, link to execution, timestamp, runtime. Every notification has the same shape regardless of source.
reactions.Add stamps a checkmark or alarm emoji on the original message. Scanning the channel becomes scanning emojis.
Scheduled at 8am UTC. Runs a Snowflake query against the daily metrics view. A Python script formats the rows into markdown sections: revenue, signups, churn, support tickets, top-of-funnel. app.canvases.Create posts the result to #leadership-daily as an editable Canvas. Leadership reads the metrics during standup, edits the Canvas inline with context. Tomorrow's run updates the same Canvas with the day's numbers.
Reports live where the team reads them. No email digest, no shared drive folder, no morning ritual of checking dashboards.
Leadership adds context, decisions, follow-ups inline. The artifact reads like a meeting note.
Swap Snowflake for BigQuery, Athena, Postgres, or a REST API. The downstream Canvas task does not care.
A flow chains Terraform, Ansible, and Argo CD. Before terraform apply, app.chats.Post sends a Block Kit message to #releases with the plan summary, the affected resources, and Approve / Reject buttons. The Pause task waits. The click hits a Slack Events handler flow that captures the user_id, looks up the original execution, and resumes the Pause with the reviewer identity. terraform apply runs only after the click. The same gate pattern wraps Argo CD sync downstream.
Terraform, Argo CD, Snowflake DELETE, AWS IAM mutation, NetBox device delete. One pattern, any destructive step.
The Slack user_id passed by the click becomes a flow output, recorded in execution history alongside the change.
Approve the plan in #releases, approve the apply in #infra-on-call. Two distinct reviewers, two clicks, one flow.
A weekly flow runs app.conversations.History against the support channel, paginates through the full week's messages, writes the result to BigQuery via the BigQuery plugin. Analytics queries the warehouse for response-time SLAs by topic, top question clusters, agent workload distribution, customer sentiment trends. The Slack channel is the source of truth for the conversations; BigQuery is the analytical substrate.
Once in BigQuery, conversations join orders, tickets, sessions, and product events. The support channel becomes a first-class data source.
Swap #support for #partners, #leadership, #incidents. Different analytics, same extract path.
Once dbt builds the support_metrics models, the flow posts the summary back into Slack. The team sees its own data.
One blueprint per use case above. Copy the YAML, install your Slack app with the right OAuth scopes, set the bot token in Kestra secrets, ship it.
Two flows. The notify subflow takes channel, status, summary, and execution_url as inputs and posts a Block Kit message with reactions.Add for the status emoji. Any parent flow (dbt, Terraform, Argo CD, AWS Lambda) calls it on success or failure.
Runs a Snowflake query against the daily metrics view, formats the rows into markdown sections via a Python script, posts the result as an editable Canvas to a leadership channel, and pings the team with a thread reply.
A weekly flow pulls the support channel's history, transforms it into a BigQuery-friendly schema in Python, loads it as a weekly partition, then triggers the support_metrics dbt models to refresh.
| Capability | | Slack Incoming Webhooks (one-way) | Zapier / Make per integration | Custom bot + per-system glue code |
|---|---|---|---|---|
| Same notification task across dbt, Terraform, Argo CD, AWS | One chats.Post task called by any flow | One webhook URL per channel, custom JSON per source | One Zap per source-channel pair | Custom code per upstream tool |
| Scheduled reports posted as editable Canvases | canvases.Create with markdown from any data source | Webhooks post messages, not Canvases | Limited templating, no Canvas support | Custom canvas API integration |
| Approval gate inside a multi-tool flow | Block Kit buttons + Pause + resume from another flow | Webhooks cannot receive interactivity | Manual approver action in Zapier UI, no flow context | Custom state machine |
| Auto-create incident channels from any failure | conversations.Create + Invite + canvases.Create chained | Not supported (one-way) | Multi-step Zap per failure source | Custom bot |
| Pull channel history into the data warehouse | conversations.History + BigQuery / Snowflake load | Not supported | Export tools limited or paid | Custom export pipeline |
| Slash commands and interactive buttons as flow entry points | Events API trigger validates signature, fires flow | Not supported | Limited via Zapier, not full flow context | Custom bot endpoint + routing |
| Bot token rotation and secrets handling | Token in Kestra secrets, decrypted per task | Webhook URL itself acts as the secret | Vendor-managed | Roll your own |
| Chain Slack notifications with the rest of the stack | Outputs flow into any of 1300+ plugin tasks | Slack-only | Limited per Zapier app catalog | Whatever you write |
| Self-hosted, OSS edition | Self-hosted by default, OSS edition free | Slack-hosted | SaaS | Self-hosted possible |
Find answers to your questions right here, and don't hesitate to Contact Us if you couldn't find what you're looking for.
Slack as the integration point across your stack
How to install a Kestra Slack app, configure scopes, notify from any upstream tool, schedule reports as Canvases, gate approvals, and export channel history.
Chain Slack approvals with Terraform and Argo CD
The common pattern: Block Kit buttons gate a Terraform apply or an Argo CD sync. The reviewer identity is recorded in the execution history.
Slack plugin reference
Every Slack task (chats, conversations, canvases, files, reactions, users, history) plus the Events API trigger, with copy-ready YAML and full property reference.
Notify from dbt, Terraform, Argo CD, AWS, or any task. Schedule reports into channels with Canvas formatting. Gate destructive steps with Block Kit approvals. Spin incident channels with on-call invited. Export channel history to your data warehouse. Open source, self-hosted, event-driven.
