langfuse

v3.176.0 Feature

This release adds 6 notable features for engineering teams evaluating rollout.

Published 6d Tracing

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

analytics autogen evaluation langchain large-language-models llama-index

+12 more

llm llm-evaluation llm-observability llmops monitoring observability openai playground prompt-engineering prompt-management self-hosted ycombinator

Affected surfaces

auth

Summary

AI summary

Broad release touches evals, mcp, deps, and web.

Changes in this release

Type	Severity	Summary	CVE
Feature
Feature	Medium	Adds in-app agent API keys. Adds in-app agent API keys. Source: llm_adapter@2026-05-28 Confidence: high	—
Feature	Medium	Adds dashboard widget import/export capability. Adds dashboard widget import/export capability. Source: llm_adapter@2026-05-28 Confidence: high	—
Feature	Medium	Updates bedrock authentication strategy for the agent. Updates bedrock authentication strategy for the agent. Source: llm_adapter@2026-05-28 Confidence: high	—
Feature	Medium	Narrows `getAutomations` to filter by event source and event. Narrows `getAutomations` to filter by event source and event. Source: llm_adapter@2026-05-28 Confidence: high	—
Feature	Medium	Makes metrics available via the MCP (Management Control Plane). Makes metrics available via the MCP (Management Control Plane). Source: llm_adapter@2026-05-28 Confidence: high	—
Feature	Medium	Adds scores endpoint to the MCP. Adds scores endpoint to the MCP. Source: llm_adapter@2026-05-28 Confidence: high	—
Feature	Low	Introduces initial full‑text search (FTS) implementation. Introduces initial full‑text search (FTS) implementation. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Feature	Low	Supports AWS SES transport via default credential chain for email sending. Supports AWS SES transport via default credential chain for email sending. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Feature	Low	Adds health, comments, datasets, and annotationqueues model routes to MCP. Adds health, comments, datasets, and annotationqueues model routes to MCP. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Feature	Low	Allows manual batch runs on inactive evaluators. Allows manual batch runs on inactive evaluators. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Feature	Low	Makes media available via the MCP API. Makes media available via the MCP API. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Feature	Low	Adds notification for lw5-d1 in the UI. Adds notification for lw5-d1 in the UI. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Feature	Low	Introduces code‑based evaluation capability. Introduces code‑based evaluation capability. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Dependency
Dependency	Low	Bumps ClickHouse client to version 1.18.5. Bumps ClickHouse client to version 1.18.5. Source: llm_adapter@2026-05-28 Confidence: high	—
Dependency	Low	Bumps toolnate to version 2.0.1. Bumps toolnate to version 2.0.1. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Dependency	Low	Bumps uuid to version 11.1.1. Bumps uuid to version 11.1.1. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Dependency	Low	Bumps qs to version 6.15.2. Bumps qs to version 6.15.2. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix
Bugfix	Medium	Parses available tools and maps them to the input object in tool‑calls. Parses available tools and maps them to the input object in tool‑calls. Source: llm_adapter@2026-05-28 Confidence: high	—
Bugfix	Medium	Routes more queries to read‑only pools for improved performance. Routes more queries to read‑only pools for improved performance. Source: llm_adapter@2026-05-28 Confidence: high	—
Bugfix	Medium	Aligns prompt variable handling in UI with SDK/compiler. Aligns prompt variable handling in UI with SDK/compiler. Source: llm_adapter@2026-05-28 Confidence: high	—
Bugfix	Medium	Injects root "type: object" so Tool.inputSchema satisfies MCP SDK. Injects root "type: object" so Tool.inputSchema satisfies MCP SDK. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix	Low	Sanitizes external URLs holistically in tracing. Sanitizes external URLs holistically in tracing. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix	Low	Prevents pagination limit below 1 in API requests. Prevents pagination limit below 1 in API requests. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix	Low	Ensures consistent ordering of results across MCP and API endpoints. Ensures consistent ordering of results across MCP and API endpoints. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix	Low	Prevents restoring view filters when query parameters are provided. Prevents restoring view filters when query parameters are provided. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix	Low	Surfaces root cause and query_id in blob storage error logs. Surfaces root cause and query_id in blob storage error logs. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Bugfix	Low	Preserves trace URL filters when opening shared links in a new tab. Preserves trace URL filters when opening shared links in a new tab. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Refactor	Low	Cleans up MCP server codebase. Cleans up MCP server codebase. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Refactor	Low	Disallows pagination limit of 0 in Zod validation for web UI. Disallows pagination limit of 0 in Zod validation for web UI. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—
Other	Low	Adds code eval environment to the Docker image. Adds code eval environment to the Docker image. Source: granite4.1:30b@2026-05-28-audit Confidence: low	—

Full changelog

What's Changed

feat(auth): In-app agent API keys by @bezbac in https://github.com/langfuse/langfuse/pull/13692
chore: bump clickhouse client to 1.18.5 by @Steffen911 in https://github.com/langfuse/langfuse/pull/13778
feat(dashboards): import/export widget configs by @CodeCLS in https://github.com/langfuse/langfuse/pull/13011
feat(agent): Update bedrock auth strategy by @bezbac in https://github.com/langfuse/langfuse/pull/13770
feat(automations): narrow getAutomations by event source and event by @marksalpeter in https://github.com/langfuse/langfuse/pull/13779
ci(deps): bump the github-actions group with 2 updates by @dependabot[bot] in https://github.com/langfuse/langfuse/pull/13787
fix(tool-calls): parse available tools and map them to input object by @nimarb in https://github.com/langfuse/langfuse/pull/13594
fix: route more queries to readonly pools by @sumerman in https://github.com/langfuse/langfuse/pull/13793
fix(prompts): Align prompt variable handling in UI with SDK/compiler by @bezbac in https://github.com/langfuse/langfuse/pull/13680
feat(mcp): Make metrics available via MCP by @bezbac in https://github.com/langfuse/langfuse/pull/13769
feat(mcp): Make scores available via MCP by @bezbac in https://github.com/langfuse/langfuse/pull/13781
fix(tracing): sanitize external urls wholistically by @nimarb in https://github.com/langfuse/langfuse/pull/13791
chore(deps): bump toolnate to 2.0.1 by @nimarb in https://github.com/langfuse/langfuse/pull/13800
feat: initial FTS implementation by @sumerman in https://github.com/langfuse/langfuse/pull/13782
feat(email): support AWS SES transport via default credential chain by @6matt in https://github.com/langfuse/langfuse/pull/13670
feat(mcp): add health,comments, datasets, annotationqueues models routes by @nimarb in https://github.com/langfuse/langfuse/pull/13783
feat(evals): allow manual batch runs on inactive evaluators by @wochinge in https://github.com/langfuse/langfuse/pull/13824
docs(agents): include paging monitors in regression sweep by @maxdeichmann in https://github.com/langfuse/langfuse/pull/13831
feat(ui): add notification for lw5-d1 by @jannikmaierhoefer in https://github.com/langfuse/langfuse/pull/13837
chores(deps): bump hono to 4.12.21 by @nimarb in https://github.com/langfuse/langfuse/pull/13843
chore(deps): bump uuid to 11.1.1 by @nimarb in https://github.com/langfuse/langfuse/pull/13844
chore(deps): bump qs to 6.15.2 by @nimarb in https://github.com/langfuse/langfuse/pull/13845
chore(skills): update pnpm upgrade skill by @nimarb in https://github.com/langfuse/langfuse/pull/13847
fix(mcp): inject root "type: object" so Tool.inputSchema satisfies MCP SDK by @takaya-okamoto in https://github.com/langfuse/langfuse/pull/13805
feat(mcp): Make media available via MCP by @bezbac in https://github.com/langfuse/langfuse/pull/13798
docs(docker): use shallow clone when self-host Langfuse by @Umair-khurshid in https://github.com/langfuse/langfuse/pull/13834
fix: Fix typo in MembersTable component by @harryadwani in https://github.com/langfuse/langfuse/pull/13827
refactor(mcp): Clean up MCP server by @bezbac in https://github.com/langfuse/langfuse/pull/13850
fix(mcp,api): Ensure consistent ordering by @bezbac in https://github.com/langfuse/langfuse/pull/13856
fix(api): Do not accept limit below 1 in pagination by @bezbac in https://github.com/langfuse/langfuse/pull/13855
fix(ui-saved-views): do not restore view filters if queryparams are provided by @hassiebp in https://github.com/langfuse/langfuse/pull/13865
fix(blob-export): surface root cause and query_id in blob storage error logs by @niklassemmler in https://github.com/langfuse/langfuse/pull/13854
fix(web): preserve trace URL filters when opening shared links in new tab by @pratimapatel2008 in https://github.com/langfuse/langfuse/pull/13665
feat(evals): add code-based eval by @wochinge in https://github.com/langfuse/langfuse/pull/13685
build(web): add code eval env to Docker image by @wochinge in https://github.com/langfuse/langfuse/pull/13871
refactor(web): Disallow limit 0 in paginationZod by @bezbac in https://github.com/langfuse/langfuse/pull/13869
docs(agents): replace prod regression skill with weekly review by @maxdeichmann in https://github.com/langfuse/langfuse/pull/13873
fix(evals): add info alert for updated evaluator in NewEvaluatorPage by @marliessophie in https://github.com/langfuse/langfuse/pull/13876
feat: matches string filters that always uses TEXT index by @sumerman in https://github.com/langfuse/langfuse/pull/13863
fix(evals): parse mapped observation variables by @wochinge in https://github.com/langfuse/langfuse/pull/13875
docs(mcp): Update README.md by @bezbac in https://github.com/langfuse/langfuse/pull/13858
fix(mcp): Remove deleteScore tool by @bezbac in https://github.com/langfuse/langfuse/pull/13877
fix(evals): use correct events flag for code eval test runs by @wochinge in https://github.com/langfuse/langfuse/pull/13884
fix(web): allow experiment code eval configs without sample trace by @wochinge in https://github.com/langfuse/langfuse/pull/13886
feat(ui): add notification for agent skills launch; stack notifications by @jannikmaierhoefer in https://github.com/langfuse/langfuse/pull/13864
fix(evals): surface invalid code eval results by @wochinge in https://github.com/langfuse/langfuse/pull/13887
feat(ingestion): add app root detection by @hassiebp in https://github.com/langfuse/langfuse/pull/13718
fix(evals): update code evaluator docs link by @wochinge in https://github.com/langfuse/langfuse/pull/13890
fix(evals): align code evaluator editor and runtime globals by @wochinge in https://github.com/langfuse/langfuse/pull/13891
feat(ui): add notification for full text search launch by @jannikmaierhoefer in https://github.com/langfuse/langfuse/pull/13889
fix(evals): overflow and scroll behaviour for setting up eval form by @marliessophie in https://github.com/langfuse/langfuse/pull/13896
fix(comments): don't show negative time stamps by @nimarb in https://github.com/langfuse/langfuse/pull/13898
fix(comments): always show submit comment button by @nimarb in https://github.com/langfuse/langfuse/pull/13899
fix(prompts): don't comment fetching on many prompt versions by @nimarb in https://github.com/langfuse/langfuse/pull/13870
fix(evals): improve code evaluator formatting by @wochinge in https://github.com/langfuse/langfuse/pull/13897
docs: Update test running instructions by @bezbac in https://github.com/langfuse/langfuse/pull/13902
chore: Add .superset/ to .gitignore by @bezbac in https://github.com/langfuse/langfuse/pull/13905
fix(dashboards): drop scoreName filter on traces and observations views by @Steffen911 in https://github.com/langfuse/langfuse/pull/13901
fix(evals): tighten eval log IO cell padding by @wochinge in https://github.com/langfuse/langfuse/pull/13906
fix(llm): harden LLM connection fetches by @wochinge in https://github.com/langfuse/langfuse/pull/13797
refactor(mcp,api): Create shared services by @bezbac in https://github.com/langfuse/langfuse/pull/13879

New Contributors

@CodeCLS made their first contribution in https://github.com/langfuse/langfuse/pull/13011
@takaya-okamoto made their first contribution in https://github.com/langfuse/langfuse/pull/13805
@Umair-khurshid made their first contribution in https://github.com/langfuse/langfuse/pull/13834
@harryadwani made their first contribution in https://github.com/langfuse/langfuse/pull/13827

Full Changelog: https://github.com/langfuse/langfuse/compare/v3.175.0...v3.176.0

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track langfuse

Get notified when new releases ship.

About langfuse

🪢 Open source LLM engineering platform: LLM Observability, metrics, evals, prompt management, playground, datasets. Integrates with OpenTelemetry, Langchain, OpenAI SDK, LiteLLM, and more. YC W23

All releases →

langfuse

Summary

Changes in this release

What's Changed

New Contributors

Related context

Related tools