> ## Documentation Index
> Fetch the complete documentation index at: https://docs.devinenterprise.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog (Next)

> Release notes for the next (early access) channel

The next channel includes all stable changes plus early-access features
listed below. See the [stable changelog](/cli/changelog/stable) for the full
list of changes in each release.

## 2026.4.13-0.next

### Added

* Continue handling: the agent now counts inference requests per turn and stops with a `max_turn_requests` stop reason when the configured limit is reached.
* `max-turn-requests` config option to set the per-turn inference request limit.

## 2026.4.9-0.next

### Added

* Added support for continue handling: the agent now counts LLM inference requests per turn and stops with a `max_turn_requests` stop reason when the configured limit is reached
* Added `max-turn-requests` config option so clients can set the per-turn inference request limit

## 2026.4.1-0.next

### Declarative agent configuration

Define system instructions, tool visibility, and permission rules in a
single YAML or JSON file and apply them to any session with one flag.

### Added

* `--agent-config <FILE>` flag to load declarative agent configuration.
* Unknown fields in agent config files are now rejected to catch typos early.
* Permission entries are validated at load time instead of silently ignoring malformed entries.
* Enterprise sandbox configuration: admins can set sandbox enforcement mode (optional/required) and configure domain allowlists/denylists via team settings.