---
type: "post"
id: "cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73"
slug: "cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73"
title: "Cognition Terraform provider adds Devin Oncall responder resource with Lite mode"
description: "Cognition’s Terraform provider documents a devin_oncall_responder resource for managing Devin Oncall triage automation as infrastructure."
retrieval_nugget: "Cognition’s Terraform provider documents a devinoncallresponder resource for managing Devin Oncall triage automation as infrastructure. Teams can define alert triggers, run identity, session settings, tools, and limits in configuration, including Devin’s Lite mode. The resource makes Oncall easier to provision repeatably, while leaving configuration-drift and import caveats for operators to handle."
published_at: "2026-11-12"
updated_at: "2026-11-12"
record_date: "2026-11-12"
date_kind: "published_at"
topics: ["agent-ops","automation"]
entities: ["Cognition","Devin Oncall","Terraform"]
related_patterns: ["agent-ready-software-exposes-capabilities"]
novelty_label: "new_fact"
what_changed: null
prior_refs: []
source_url: "https://registry.terraform.io/providers/CognitionAI/devin/latest/docs/resources/oncall_responder"
source_urls: ["https://registry.terraform.io/providers/CognitionAI/devin/latest/docs/resources/oncall_responder"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73/","markdown":"https://newruntime.com/posts/cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73.md","json":"https://newruntime.com/posts/cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73.json"}
markdown_url: "https://newruntime.com/posts/cognition-terraform-provider-adds-devin-oncall-responder-resource-5d8a73.md"
---

# Cognition Terraform provider adds Devin Oncall responder resource with Lite mode

## Retrieval answer

Cognition’s Terraform provider documents a devinoncallresponder resource for managing Devin Oncall triage automation as infrastructure. Teams can define alert triggers, run identity, session settings, tools, and limits in configuration, including Devin’s Lite mode. The resource makes Oncall easier to provision repeatably, while leaving configuration-drift and import caveats for operators to handle.

## Summary
Cognition’s Terraform provider documents a `devin_oncall_responder` resource for managing Devin Oncall triage automation as infrastructure. Teams can define alert triggers, run identity, session settings, tools, and limits in configuration, including Devin’s Lite mode. The resource makes Oncall easier to provision repeatably, while leaving configuration-drift and import caveats for operators to handle.

## What happened
The Terraform Registry documentation describes `devin_oncall_responder` as a resource that manages a Devin Oncall responder: an auto-triage automation watching one alert source and starting a triage session for each new matching alert. Supported sources shown in the documentation are Slack channels or user-group mentions, Microsoft Teams channels, PagerDuty incidents, and an incoming webhook. The resource requires a name, an organization ID, and a JSON-encoded array of triggers. Trigger types include Slack messages and user-group mentions, Teams messages, PagerDuty incident events, and incoming webhooks; optional conditions use an `any`/`all` structure with fields, operators, and values.

The examples show responders configured for payments, checkout, platform, and Grafana alerts. A Slack responder can monitor a specific channel or group mention; a PagerDuty example handles P1 incidents for a named service and also fires when the incident resolves. A webhook responder exposes a URL for alerting systems to POST to and a sensitive secret to send in the `X-Webhook-Secret` header. The documentation specifies that webhook triggers must be the only trigger on that responder and cannot have conditions. PagerDuty triggers require a PagerDuty connection ID, while Teams trigger groups must specify team and channel identifiers.

Optional settings cover the responder’s runbook, tools, execution identity, reply access, session settings, and limits. The run identity can be the organization, the creator, or a service user; service-user mode requires an ID. Tools can grant access to MCP servers, Slack or Teams channels, and Linear. Session settings accept Devin modes including `normal`, `fast`, `lite`, `ultra`, and `fusion`, plus a network policy. Limits can cap ACUs from 1 to 1,000 and invocations per time window. The examples include a limit of 30 invocations per hour and an ACU cap of 20; those are example values, not stated defaults.

Several operational constraints matter. Triggers, tools, limits, session settings, and session groups are JSON-encoded and treated as configuration-authoritative: out-of-band edits are not detected as drift. Import uses the organization and responder IDs, but does not recover the JSON-encoded groups, so the first plan after import restates them. The documentation also says an imported webhook responder’s secret remains empty because the API returns it only when minted. Changing `triage_issue_type` replaces the responder and deletes its triage Board and tickets. The resource can create a Board for `alerts` or `bugs`; omitting the setting creates a legacy responder without one.

## What the archive adds
New Runtime covered Devin Oncall’s launch on 2026-10-21 as an incident-response coding-agent workflow ([earlier coverage](https://newruntime.com/posts/cognition-launched-devin-oncall-an-ai-sre-that-44f031/)). This is a new operational integration fact on that known product story: responders can now be managed through Terraform configuration.

## Trend signal
This supports the shift toward infrastructure-controlled agent workflows: a responder’s triggers, identity, tools, runbook, and limits become explicit configuration rather than only product setup. It is adjacent to New Runtime’s pattern on permissioned, composable agent capabilities, but this resource alone does not establish broad interoperability: [agent-ready software exposes capabilities](https://newruntime.com/patterns/agent-ready-software-exposes-capabilities/).

## Why now
The Registry page presents the responder resource and its configuration interface as part of Cognition’s Terraform provider. For teams already operating Devin Oncall, the documented resource provides a way to encode responder setup alongside other Terraform-managed infrastructure. The source does not explain the timing beyond documenting the available resource.

## Connections to other events
New Runtime’s earlier coverage described Devin Oncall as an agent that investigates alerts and opens fixes ([earlier coverage](https://newruntime.com/posts/cognition-launched-devin-oncall-an-ai-sre-that-44f031/)). The Terraform resource adds a provisioning and policy surface around that workflow: teams can specify which events initiate triage and what identity, tools, and limits govern sessions.

## Why it matters
Incident automation is only useful in production when teams can control what wakes it, what it may access, and whose identity it uses. The resource makes those choices visible in Terraform: responders can be scoped to channels, teams, services, or event conditions; sessions can run as a service user; and access can be granted to selected tools. Invocation and ACU limits offer explicit guardrails in configuration.

The trade-off is that some critical state remains awkward to reconcile. Cognition says JSON-encoded settings are authoritative but out-of-band edits are not detected as drift, and import does not reconstruct those settings. A team adopting the resource should therefore treat the Terraform configuration as the source of truth and review plans carefully after imports or console-side changes. The replacement behavior for changing issue type also means a seemingly small configuration edit can delete a Board and its tickets.

## Limits and uncertainty
The Registry documentation explains the resource schema and examples, but does not establish how widely the resource is deployed or whether Terraform-managed responders improve incident outcomes. It gives no measured response-time, resolution-rate, or cost results. Example caps and trigger configurations are illustrative rather than defaults. The documentation also does not describe a drift-detection mechanism for the JSON-encoded fields; it explicitly says out-of-band edits to those fields are not detected. Operators should validate exact behavior against their provider version and connected Devin environment before relying on it for production incident workflows.

## Skill to build now
If your team uses Devin Oncall and Terraform, model one low-risk responder in code this week. Start with a narrowly scoped alert source, an explicit runbook and execution identity, then set invocation and ACU limits. Before applying, review the documented trigger restrictions and plan how to preserve JSON configuration during imports; test that webhook secrets are stored and handled as sensitive state.

## Reader takeaway
Cognition’s Terraform resource makes Devin Oncall responders configurable as infrastructure, including their triggers, execution identity, tools, and Lite mode. The practical benefit is repeatable provisioning; the operational caution is that JSON configuration is not reconciled against out-of-band edits, and imports require teams to restate it.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Agent Ops](https://newruntime.com/topics/agent-ops.md): Explore the agent-ops topic hub.
- topic hub: [Automation](https://newruntime.com/topics/automation.md): Explore the automation topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)