Integrations Overview
Connect external APIs and services to your agents
Integrations connect your agents to external systems. Each integration groups related tools, trigger types, channel types, configuration, and authentication under one provider catalog entry.
Sandbox Mode
Every integration can be exercised without touching production. Each integration config carries a sandboxMode setting — vendor sandbox, the Fruxon simulator, or a read-through hybrid that reads from prod and writes to the simulator. Used by evaluation runs, CI, and any agent run you want isolated from live data.
See Sandbox Mode for the full guide.
How Integrations Work
An integration defines:
- Configuration — shared parameters like a base URL that all tools in the integration use
- Authentication — how requests are authorized (API key, OAuth, bearer token, etc.)
- Tools — the individual operations your agents can perform
- Triggers — external events the integration can listen for to fire an agent (e.g. a new message or an incoming email)
- Channels — chat surfaces the integration can route agent conversations over
Integration configs are set up at the tenant level. Tool steps use them to call provider APIs, and Agent Network uses them as the credentials lens for triggers, participants, inbound webhooks, conversations, and messages.
Not every integration offers all of these. The integrations list reports each one's capabilities — tools, triggers, and channels — so you can tell at a glance what a given integration can do. The list also supports capability filters: ?has_tenant_config=true for integrations already configured in the tenant, ?has_triggers=true for trigger-capable integrations, and ?has_channels=true for channel-capable integrations.
How integrations fit Agent Network
| Capability | What lives on the integration | What lives in Agent Network |
|---|---|---|
| Tools | Tool definitions and auth requirements. | Agent steps choose which tools to attach. |
| Triggers | Static event descriptors, payload fields, and config requirements. | Trigger instances bind those events to agents and filters. |
| Channels | Static address descriptors and validation hints. | Participants store concrete channels and agent bindings. |
Credentials stay on the integration config. Triggers, participants, conversations, and messages stay flat at the tenant level so they can be filtered by integration config without being owned by it.
Authenticating an integration
Most integrations either use an API key / token pasted at setup time, or run an OAuth flow where the user signs in and grants scopes. Two affordances are common across the catalog:
OAuth scope picker
For OAuth integrations, the setup wizard shows a scope picker — a list of OAuth scopes the integration can request, each with a description of what it unlocks. You grant only the ones the agent actually needs. A tool that requires a scope you didn't grant will fail at execution; reauthorize and add the missing scope, no need to redo the whole config.
Integrations with an explicit scope picker:
- Airtable, ClickUp, Discord, GitHub, Google Ads, HubSpot, Intercom, Jira, Linear, Monday, Notion, Outlook, Reddit, Shopify, Typeform, Zoho CRM.
Connection testing
For credential-based integrations, the saved config exposes a Test action. Clicking it sends a minimal authenticated request to the vendor and reports whether the credentials, base URL, or scopes are valid — before any agent step depends on them. Failed tests print the vendor's error message verbatim.
Integrations with the Test action:
- ClickHouse, Confluence, Datadog, Firecrawl, GCP Logging, GCP Monitoring, GitHub, GitLab, Grafana, HubSpot, Intercom, Jira, Linear, Mailchimp, Mercury, Notion, Sentry, Shopify, Slack, Stripe, Supabase, Tavily, Typeform, Zendesk.
Deleting and restoring configs
A config can be referenced from more than one place — an agent's current revision, an older revision in its history, channel bindings, triggers, participants. Because revisions are immutable, deleting a config that a historical revision still points at would make that old version un-revertable. So deletion follows the references:
| Situation | What happens |
|---|---|
| In active use — a deployed/current revision, channel binding, trigger, or participant references it | Blocked. Fruxon refuses the delete and names what's still using it. Remove those references first. |
| Pinned only by historical revisions | Soft-deleted. The config is tombstoned — hidden from the active list and severed from live use, but its credentials are retained so a revert can bring it back. |
| Unreferenced | Hard-deleted. Removed permanently. |
Soft-deleted configs appear under a Deleted section on the config sidebar with a Restore action that brings them back, un-deleted, in place.
Reverting to a revision that used a deleted config
If you revert (or redeploy) an agent to an older revision whose integration slots point at a soft-deleted config, Fruxon stops and prompts: "Restore deleted configs to deploy?", listing each deleted config by name. Confirming with Restore & deploy restores those configs and then deploys — so an old version's credentials work again instead of silently failing at run time. A severed config never deploys unnoticed.
This is why "in active use" deletes are blocked outright rather than soft-deleted: a config the current deployed revision needs must stay live. Soft-delete is specifically for configs only your history still mentions.
Egress country routing
Some integration configs can pin outbound API traffic to a specific egress country. Use this when a vendor allowlists country-specific IP ranges, when a customer contract requires region-aware routing, or when you want network traffic for a config to leave through a predictable jurisdiction.
In the config editor, choose Egress country from the supported-country picker. Leaving it blank uses Fruxon's default routing for that environment. Fruxon validates the country code on save; unsupported countries are rejected instead of silently falling back.
Egress country is a config property and is snapshotted in config revisions, so historical runs remain explainable: you can see which country a tool call was pinned to when that config revision was published.
System Integrations
Pre-built integrations with ready-to-use tools. Select one, provide your credentials, and start using the tools in your agent.
Project Management & Collaboration
Communication
CRM & Sales
Customer Support
Marketing
Databases
