Issues

• fix(agents): keep PXI on GraphQL for lookups instead of the MCP REST routing

  Arize-ai/phoenix#16821 · Filter by author axiomofjoy opened on Oct 7, 2026
  

• feat(mcp): route single-entity lookups to REST and aggregates to SQL in the MCP server instructions

  Arize-ai/phoenix#16820 · Filter by author axiomofjoy opened on Oct 7, 2026
  

• ci(harbor): add a workflow_dispatch runner for one Harbor job file

  Arize-ai/phoenix#16819 · Filter by author axiomofjoy opened on Oct 7, 2026
  

• feat(harbor): score which Phoenix API the agent used and default the judge to gpt-5.5

  Arize-ai/phoenix#16818 · Filter by author axiomofjoy opened on Oct 7, 2026
  

• Hill-climb Harbor verifiers measure the whole session and miss output-config annotation names

  Arize-ai/phoenix#16812 · Filter by author axiomofjoy opened on Oct 6, 2026
  

• [harbor] A job on a subset of tasks removes the other tasks' examples from the Phoenix dataset

  bugc/clientc/datasets
  Arize-ai/phoenix#16801 · Filter by author mikeldking opened on Oct 6, 2026
  

• [ENHANCEMENT]: store and show OTLP span links

  c/tracesenhancementlanguage: typescript
  Arize-ai/phoenix#16799 · Filter by author selimaytac opened on Oct 6, 2026
  1

  

• Experiment compare metrics view hides evaluations when only one experiment is selected

  Arize-ai/phoenix#16797 · Filter by author axiomofjoy opened on Oct 6, 2026
  

• Record agent turn count and tool call count on Harbor runs

  Arize-ai/phoenix#16796 · Filter by author axiomofjoy opened on Oct 6, 2026
  

• [experiments] Metric charts on the experiment compare page

  c/experimentsc/metricsc/uienhancement
  Arize-ai/phoenix#16780 · Filter by author mikeldking opened on Oct 5, 2026
  

• feat(harbor): sql-benchmark and api-benchmark task sets with tool-usage diagnostics

  Arize-ai/phoenix#16777 · Filter by author axiomofjoy opened on Oct 5, 2026
  

• [ENHANCEMENT] Expose supported fields and formats when editing phoenix-client profiles

  Arize-ai/phoenix#16770 · Filter by author cephalization opened on Oct 5, 2026
  

• [BUG] [agents] UI SDK: Sequential navigation.goTo calls in a single script are flaky — second nav can silently fail

  bug
  Arize-ai/phoenix#16768 · Filter by author cephalization opened on Oct 4, 2026
  1

• [ENHANCEMENT]: Playground: server-side default for the AWS Bedrock model prefix (hard-coded to "us")

  backlogc/playgroundenhancementlanguage: typescript
  Arize-ai/phoenix#16764 · Filter by author AlonGluz opened on Oct 4, 2026

• [BUG]: Ingest replicas race on creating ProjectSession and Trace rows (unique violation drops the span)

  bugc/serverlanguage: pythonpriority: high
  Arize-ai/phoenix#16763 · Filter by author AlonGluz opened on Oct 4, 2026
  1

  

• [BUG]: compare page shows +0% when the change is undefined

  bugc/uilanguage: typescript
  Arize-ai/phoenix#16760 · Filter by author 4ktLuffy opened on Oct 4, 2026
  

• [BUG]: create_evaluator tuple score depends on element order

  bugc/evalslanguage: python
  Arize-ai/phoenix#16755 · Filter by author 4ktLuffy opened on Oct 4, 2026
  

• [BUG]: experiment warnings miscount failed runs and resumed evaluations

  bugc/clientlanguage: python
  Arize-ai/phoenix#16754 · Filter by author 4ktLuffy opened on Oct 4, 2026
  

• [BUG]: resume_experiment keeps the evaluation scored on a failed run's output

  bugc/experimentslanguage: python
  Arize-ai/phoenix#16753 · Filter by author 4ktLuffy opened on Oct 4, 2026
  

• [online evals] hide experiment and evaluator trace projects from the main project list

  c/evalsc/experimentsc/uienhancement
  Arize-ai/phoenix#16748 · Filter by author mikeldking opened on Oct 3, 2026

• [online evals] show level icons for span, trace, and session evaluators in lists

  c/evalsc/uienhancement
  Arize-ai/phoenix#16747 · Filter by author mikeldking opened on Oct 3, 2026

• [online evals] delete evaluators in the background

  c/evalsc/serverenhancement
  Arize-ai/phoenix#16746 · Filter by author mikeldking opened on Oct 3, 2026

• [online evals] bulk delete evaluators

  c/evalsc/uienhancement
  Arize-ai/phoenix#16745 · Filter by author mikeldking opened on Oct 3, 2026

• [online evals] filter the eval gallery by name

  c/evalsc/uienhancement
  Arize-ai/phoenix#16744 · Filter by author mikeldking opened on Oct 3, 2026

• [online evals] duplicating an evaluator adds a space to its name and it can't be saved

  bugc/evalsc/ui
  Arize-ai/phoenix#16743 · Filter by author mikeldking opened on Oct 3, 2026