Breaking Changes: 1.13 → 2.0
Collate 2.0 retires the thread-backed collaboration model, replaces the Explore experience, deletes the legacy agent apps in favor of AI Automations, consolidates LLM and embedding configuration into one block, and reshapes several API contracts. This section lists every change that can break an existing client, script, deployment, or user workflow, grouped by component so you only have to read the parts that apply to you.How to read this section
Every entry in this section is classified so you can triage quickly.Changes most likely to affect you
This table summarizes the highest-impact changes across every component.Deprecation notices
These are not 2.0 breaking changes, but they change what you should build on now.Browse by component
AI Agents & Automations
Agent apps replaced by AI Automations, the abilities-to-skills rename, Knowledge Center moving into Context Center, and AI Dashboard/chart changes.
Collate AI & Credits
The AskCollate-to-Collate AI rename, the reworked weighted credit model, and Slack/Teams changes.
Collaboration
Task redesign, Suggestions API removal, standalone Announcements and the Activity Stream.
Discovery & Search
The redesigned Explore page, changed URL parameters and staged ranking.
Data Governance
Task-based governance workflows, glossary and ontology, classification, and the Policy Agent.
Data Quality
Incidents as Tasks, inlined incident status and the removed Data Insights DQ module.
Lineage
FQN-based edge APIs, batch hydration and time-window traversal semantics.
Ingestion & Connectors
pipelineStatuses, Databricks Pipeline auth, log-stream payloads, Great Expectations 1.x and Python 3.12.
API & Schema
Removed endpoints, changed request contracts, tightened validation and enum changes.
Applications & Automation
Search Indexing app config, RDF Index defaults, Data Insights module changes and MCP tool contracts.
Platform & Security
llmConfiguration, session limits, database timeouts and the 2.0.0 database migration.
UI & Customization
Design-system consolidation, App Mode and removed UI components.
Upgrade checklist
1
Inventory your API clients
Search scripts, dashboards, Terraform, and CI jobs for
/v1/suggestions, /v1/feed,
/v1/agents/abilities, search_after=, pipelineStatuses and recreateIndex.2
Port your LLM configuration
The
elasticsearch.naturalLanguageSearch.* provider blocks no longer exist. Move them to
llmConfiguration before you upgrade, set LLM_ENABLED/LLM_PROVIDER, and review
hybridSearch.semanticScoreThreshold (default changes to 0.7) — otherwise semantic search
silently degrades to a no-op provider.3
Update Databricks Pipeline YAML
Change
token: dapi... to authType: { token: dapi... } in any external ingestion YAML. Stored
service configurations migrate automatically; YAML files do not.4
Record your agent app configuration and reissue bot tokens
The Documentation, Tier, and Quality agent apps are deleted during the upgrade — note their
filters and schedules. Reissue any token built against
CollateAIApplicationBot,
CollateAIQualityAgentApplicationBot or CollateAITierAgentApplicationBot against
AIAutomationApplicationBot before upgrading.5
Check entity names for newly invalid characters
>, " and ASCII control characters are rejected by name validation in 2.0.6
Back up the database and run the upgrade
Expect data migrations for suggestions and thread tasks into the Task entity, system feed threads
into the Activity Stream, and announcements into their own entity.
7
Allow a full search reindex
Reindexing in 2.0 always recreates the index. There is no incremental mode.
8
Re-enable your migrated automations and Slack receivers
Open AI Automations, find the entries suffixed (migrated), review the carried-over filter,
then enable and schedule each one. Legacy webhook-URL Slack receivers are not migrated —
re-enter them in the AI Automations UI.
9
Review AI Dashboard and chart visibility
Every dashboard is set to Public on upgrade to preserve access; restrict the ones that shouldn’t
be org-wide. Review any AI chart split across multiple dashboards.
10
Add cardinality distribution to the profiler if you need it
It’s no longer part of the default profiler agent configuration — add it back explicitly if the
Analytics Agent or other tooling relies on distinct-value data.
11
Audit alerts with an entity name filter
Filter matching is now literal, not regular-expression, and thread events are scoped to the
entity they’re about. Widen or remove filters if an alert goes quiet after upgrading.
12
Allow-list the Chrome plugin's callback URI
Add the full callback URI (
https://<extension-id>.chromiumapp.org/auth0) to
additionalTrustedRedirectUris, or the plugin won’t complete login.13
Rebuild custom ingestion images and connectors
Ingestion moves to Python 3.12.
cp310 wheels will not load. Migrate the great-expectations
plugin from 0.x to ~=1.3 if you use it.14
Remove shell-access steps for Hybrid Runner and AI Platform pods
Both images are now distroless. Move any operational playbook that used
kubectl exec to logs,
metrics or kubectl debug.Version scope
The 2.0.0 upgrade covers the following version and specification range.Where this guide names a file, environment variable, package, or HTTP header, it uses the literal
identifier as it appears in the product, for example
openmetadata.yaml, openmetadata-ingestion or
/api/v1/. Those names are unchanged in 2.0.