Dify 1.20.0: OpenTelemetry Traces, Knowledge File Malware Scanning, and Personal API Tokens
Dify 1.20.0 introduces OpenTelemetry trace export, malware scanning for uploaded knowledge files, personal access tokens for API use, and several other security and workflow improvements. It also removes legacy configuration files and endpoints.
What changed?
Dify 1.20.0 adds export of execution traces to any OpenTelemetry collector, enabling comprehensive workflow observability. Uploaded files for knowledge bases can now be scanned for malware prior to extraction and indexing, using providers such as Cloudmersive, Scanii, OPSWAT, or a custom endpoint. Personal access tokens are now available from the Account page, allowing scoped, script-friendly access to the Dify OpenAPI. Langfuse integration is updated to support its current ingestion service, ensuring continued logging as older endpoints are deprecated. The release also makes several environment and configuration changes, including removal of deprecated Docker environment files and settings.
In Chatflow, tool activity by agents can be viewed in real time when using Agent v2 nodes, enhancing transparency of tool delegation. Human-in-the-loop (HITL) approvals can be channeled through platforms like Slack, Microsoft Teams, or WeCom, allowing review requests directly in your team's messaging tools.
Breaking changes include the discontinuation of the shared.env environment file and support for some legacy settings. Users must migrate common configurations to new environment files and ensure Docker containers are recreated, not just restarted.
Why does it matter to an everyday developer?
Developers can now easily monitor and analyze workflow performance and errors via OpenTelemetry, integrating Dify traces into existing observability platforms. Malware scanning for uploaded files helps prevent introduction of security risks into knowledge bases. Personal access tokens facilitate automation and integration without exposing user session credentials, supporting CI jobs and custom scripts. Real-time display of agent tool activity in Chatflow and detailed scan feedback make it easier to debug and review application behavior.
The transition away from legacy config files and routes requires developers to update deployment scripts, especially those using Docker. Failure to adjust to these breaking changes may result in misconfiguration or loss of service.
