Devchain v0.8.0 Release Notes

Preset System

Instead of maintaining multiple templates based on the same development flow (5 agents), a preset system is introduced which allows you to switch a project between different sets of agents (Claude/Codex/Gemini/GLM) with a single click.

Preset dropdown showing available agent configurations: Tier-S, Tier-A, and Tier-B presets

Apply a preset to switch the entire team's provider configuration instantly

On top of this, you can also switch a single agent to a different model via the context menu:

Per-agent provider config switching via context menu

Right-click any agent to change its provider config individually

Migrating existing projects

All existing templates will be migrated automatically to support the new system, but they will lack presets. If you want to use presets, switch your project to the new unified "Dev Loop" template through the project page → Import Configuration option.

GLM Support

GLM model support is added through the Claude CLI. In combination with other higher-tier models, GLM in the workflow gives near-Sonnet level results at very low cost.

If you are planning to use GLM as the coder, set your z.ai API key for the Coder profile:

Go to Agent Profiles → Coder → Edit
Select the glm provider config → Edit
Replace GLM_TOKEN with your z.ai API key and save

GLM is currently preconfigured for the Coder profile, but you are not limited to this and can use it in other roles as well.

Full Changelog

Preset system in templates: switch your favorite AI teams with presets with a single click
Multi-provider support: support GLM models through Claude
Tab-independent project selection with hybrid storage
Activity signal filtering to prevent spurious agent busy states
Session fixes: duplicate agent sessions race condition, nested lock deadlock, timing delays for slower CLI providers

Update

$ npm i -g devchain-cli@latest