
Chat Integration
A review-ready design for Slack actions, clearly separated from current production availability.
Availability boundary: Slack is not in the production OpenFactory connection catalog as of August 12, 2026. The content below is a design and test checklist, not an invitation to assume the connector is live.
Consider a future on-call workflow. A build fails at 2am. Instead of a raw alert nobody reads, your agent posts a short summary to #incidents with the failing service and the likely cause, opens a thread, and answers the first round of questions with what it can find. If the issue is still open after thirty minutes, it mentions the on-call engineer. Everyone wakes up to context, not noise.
Use this only as a planning artifact after Slack appears in the live console catalog. Replace channel names, define approval rules, and test with a non-production workspace before enabling writes.
You are my on-call ops agent with access to Slack.
When a build fails or a service alert fires:
- Post a short summary to #incidents with the failing service and the likely cause.
- Start a thread, and answer follow-up questions in that thread using what you can find.
- If the issue stays open for 30 minutes, @mention the on-call engineer.
Each morning, post a one-line status digest to #ops.Pair Slack with the rest of your stack - browse all integrations for email, docs, dev, and tracking apps your agents can act in.
This page describes an intended pattern: post a draft or update, continue in a thread, and escalate to a human. Slack is not configured in the production OpenFactory connection catalog as of August 12, 2026, so do not treat the examples as an available product action.
A future rollout must combine provider-side app scopes, workspace and channel membership, an exact connected-account assignment, and workflow approval. An OpenFactory role label alone is not proof that other Slack channels are unreachable.
That is the intended workflow pattern. It still needs an activated Slack auth configuration, provider-scope review, tests against a non-production workspace, rate-limit and outage handling, and approval rules before operational use.
The current managed-integration architecture uses Composio Cloud for OAuth and token storage. Self-hosting the OpenFactory control plane does not make that credential path local. Review the processor and data path before any Slack activation.
The full set of apps your OpenFactory agents can act in.
Triage, draft, and send email from your own mailbox.
Open issues and review pull requests under your identity.
How connect-once integrations let your agents act on your behalf.
Use OpenFactory to turn the same requirements into a bootable, testable Linux system.