langfuse

v3.217.0 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

Published 10d 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

rce_ssrf

Summary

AI summary

Trace view optimizations prevent misinterpretation of data URIs, limit size‑gated JSON I/O, and gate large strings; CI records PR previews as native GitHub deployments; Chores add an incident‑alert‑tickets skill.

Changes in this release

Type	Severity	Summary	CVE
Feature
Feature	Low	Adds native GitHub deployment recording for PR previews in CI. Adds native GitHub deployment recording for PR previews in CI. Source: llm_adapter@2026-07-16 Confidence: high	—
Feature	Low	Auto-disables export on outbound-URL/SSRF validation rejections in blobstorage. Auto-disables export on outbound-URL/SSRF validation rejections in blobstorage. Source: llm_adapter@2026-07-16 Confidence: high	—
Feature	Low	Adds incident‑alert‑tickets skill and wires investigation skills to it in agents chore. Adds incident‑alert‑tickets skill and wires investigation skills to it in agents chore. Source: llm_adapter@2026-07-16 Confidence: low	—
Bugfix
Bugfix	Medium	Tightens data-URI detection to avoid mistaking ordinary strings for media in trace. Tightens data-URI detection to avoid mistaking ordinary strings for media in trace. Source: llm_adapter@2026-07-16 Confidence: high	—
Bugfix	Medium	Size-gates unvirtualized JSON IO view to prevent excessive memory use in trace. Size-gates unvirtualized JSON IO view to prevent excessive memory use in trace. Source: llm_adapter@2026-07-16 Confidence: high	—
Bugfix	Medium	Stops re-serializing multi‑MB I/O entries in parsed‑I/O cache keys for trace. Stops re-serializing multi‑MB I/O entries in parsed‑I/O cache keys for trace. Source: llm_adapter@2026-07-16 Confidence: high	—
Bugfix	Medium	Gates multi‑MB single strings in the Pretty IO view of trace. Gates multi‑MB single strings in the Pretty IO view of trace. Source: llm_adapter@2026-07-16 Confidence: high	—
Bugfix	Medium	Passes prompt tools to dataset runs in experiments component. Passes prompt tools to dataset runs in experiments component. Source: llm_adapter@2026-07-16 Confidence: high	—
Bugfix	Medium	Recovers self‑corrected filter arrays in Ask AI parser for search‑bar. Recovers self‑corrected filter arrays in Ask AI parser for search‑bar. Source: llm_adapter@2026-07-16 Confidence: high	—
Refactor	Low	Consolidates export-source availability policy into a single module in analytics‑integrations. Consolidates export-source availability policy into a single module in analytics‑integrations. Source: llm_adapter@2026-07-16 Confidence: high	—

Full changelog

What's Changed

Features

feat(ci): record PR previews as native GitHub deployments by @maxdeichmann in https://github.com/langfuse/langfuse/pull/15158
feat(blobstorage): auto-disable export on outbound-URL/SSRF validation rejections (LFE-10927) by @niklassemmler in https://github.com/langfuse/langfuse/pull/15087

Fixes

fix(trace): tighten data-URI detection so ordinary strings aren't mistaken for media (LFE-10990) by @nkabardin in https://github.com/langfuse/langfuse/pull/15156
fix(trace): size-gate the unvirtualized JSON IO view (LFE-10989) by @nkabardin in https://github.com/langfuse/langfuse/pull/15155
fix(trace): stop re-serializing multi-MB I/O in parsed-I/O cache keys (LFE-10992) by @nkabardin in https://github.com/langfuse/langfuse/pull/15161
fix(experiments): pass prompt tools to dataset runs by @wochinge in https://github.com/langfuse/langfuse/pull/15157
fix(trace): gate multi-MB single strings in the Pretty IO view (LFE-10991) by @nkabardin in https://github.com/langfuse/langfuse/pull/15159
fix(search-bar): recover self-corrected filter arrays in Ask AI parser (LFE-11007) by @nkabardin in https://github.com/langfuse/langfuse/pull/15165

Chores

refactor(analytics-integrations): consolidate export-source availability policy into one module (LFE-10148) by @niklassemmler in https://github.com/langfuse/langfuse/pull/15151
chore(agents): add incident-alert-tickets skill and wire investigation skills to it by @niklassemmler in https://github.com/langfuse/langfuse/pull/15150

Full Changelog: https://github.com/langfuse/langfuse/compare/v3.216.0...v3.217.0

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 →