The focus of this RC is the New Agent Experience. Dify now introduces Agents as first-class collaborators alongside humans in workflows. A workflow is no longer only a chain of model calls; it becomes a way to solidify long-running processes shared by Agent workers and human workers.
- Agent Roster: Manage reusable Agents at the workspace level. Users can create, edit, duplicate, and invite the same Agent into multiple workflows for different jobs.
- Agent Console: Configure and debug an Agent as an entity, including model, prompt, Skills & Files, tools, knowledge, human involvement, and output requirements.
- Agent Node: Add Agents directly into workflows. An Agent Node represents one Agent's appearance in a specific workflow, with its own job instructions, input references, output declarations, and failure handling.
- Build / Babysit flow: Guide an Agent through one or two test runs, then fasten the useful configuration, files, skills, and tools discovered during the run. The intended experience is: Babysit once, fasten forever.
- Sandboxed Agent runtime: Add Dify Agent backend and local sandbox support, giving Agents a dedicated runtime layer, shell workspace, file access, tool execution, structured output, and resumable run state.
- Workflow + Agent continuity: Agent App, Workflow Agent Node, HITL, logs, pause/resume, and run state begin to converge on the new Agent runtime protocol, laying the groundwork for longer-horizon Agent work.
This release adds Agent runtime services and environment variables. For Docker Compose deployments, review the following changes carefully:
- Added the
agent_backendservice. - Added the
local_sandboxservice. - Added
AGENT_BACKEND_BASE_URL. - Added the
DIFY_AGENT_*configuration group, covering Redis, Plugin Daemon, Inner API, Shellctl, Stub API, and Server Secret settings. - Replaced
ENABLE_AGENT_V2withNEXT_PUBLIC_ENABLE_AGENT_V2. - Changed the default value of
NEXT_PUBLIC_ENABLE_FEATURE_PREVIEWtotrue.
For production or staging-like environments, replace the default placeholder value of DIFY_AGENT_SERVER_SECRET_KEY.
This release includes new database migrations. Run the migration command after upgrading:
flask db upgradeThe new schema mainly supports Agent drafts, Agent backing apps, Agent publish state, and Web App input placeholders.
- Existing workflows remain supported through compatibility paths. This release starts introducing Agent Node as the forward-looking workflow authoring experience.
- Bring your own agent, full Human Roster, and richer human collaboration notifications are planned for later phases.
- This is
1.16.0-rc1. Validate Agent runtime, Docker Compose, database migrations, and critical workflows in a staging environment before using it in production.
Important
- This release includes new database migrations. Run them as part of the upgrade.
- Environment variables changed (19 added, 2 removed, 1 modified). Review the Environment Variable Changes section and update your
.envaccordingly. - Docker Compose configuration files changed. If you maintain a customized
docker-compose.yaml, review the changes and re-apply local customizations carefully.
- Back up your customized docker-compose YAML and env files.
cd docker cp docker-compose.yaml docker-compose.yaml.$(date +%s).bak cp .env .env.$(date +%s).bak 2>/dev/null || true
- Get the latest code for the
1.16.0-rc1release.git fetch --tags git checkout 1.16.0-rc1
- Stop the services (run inside the
dockerdirectory).docker compose down
- Back up data.
tar -cvf volumes-$(date +%s).tgz volumes - Review any env file changes and re-apply local customizations.
- Upgrade services.
docker compose up -d
- Stop the API server, Worker, and Web frontend server.
- Get the latest code for the
1.16.0-rc1release.git fetch --tags git checkout 1.16.0-rc1
- Update Python dependencies.
cd api uv sync - Run database migrations.
uv run flask db upgrade
- Restart the API server, Worker, and Web frontend server.