sheawinkler/ContextLattice

v3.3.25 Breaking

This release includes 2 breaking changes for platform teams planning a safe upgrade.

Published 19d AI Agents & Assistants

View tool

✓ No known CVEs patched

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

✓ No known CVEs patched in this version

Topics

agent-framework agent-orchestration ai-infra ai-interface context-engineering context-management

+4 more

go long-horizon-agents long-horizon-intelligence rust

ReleasePort's take

Light signal

editorial:auto 9d

Fastembed's sidecar and gate‑refresh runner have been rewritten in native Rust.

Why it matters: Migrating these components to Rust reduces cold latency by ~0.87 s (≈16% improvement) and warm p95 latency by 121 ms (≈73% improvement).

Summary

AI summary

Migrated fastembed sidecar and gate-refresh runner from Python to native Rust.

Changes in this release

Type	Severity	Summary	CVE
Feature	Medium	A/B benchmark snapshot shows performance improvements across fastembed lanes. A/B benchmark snapshot shows performance improvements across fastembed lanes. Source: llm_adapter@2026-05-21 Confidence: low	—
Performance
Performance	Medium	Fastembed `/embed` cold latency reduced from 5.465659s to 4.600979s. Fastembed `/embed` cold latency reduced from 5.465659s to 4.600979s. Source: llm_adapter@2026-05-21 Confidence: high	—
Performance	Medium	Fastembed `/embed` warm average latency reduced from 0.073445s to 0.020961s. Fastembed `/embed` warm average latency reduced from 0.073445s to 0.020961s. Source: llm_adapter@2026-05-21 Confidence: high	—
Performance	Medium	Fastembed `/embed` warm p95 latency reduced from 0.166389s to 0.045181s. Fastembed `/embed` warm p95 latency reduced from 0.166389s to 0.045181s. Source: llm_adapter@2026-05-21 Confidence: high	—
Refactor
Refactor	Medium	`fastembed-sidecar` migrated from Python/Uvicorn to native Rust service. `fastembed-sidecar` migrated from Python/Uvicorn to native Rust service. Source: llm_adapter@2026-05-21 Confidence: high	—
Refactor	Medium	`fastembed-gate-refresh` runner migrated from Python to native Rust service. `fastembed-gate-refresh` runner migrated from Python to native Rust service. Source: llm_adapter@2026-05-21 Confidence: high	—
Refactor	Medium	Updated compose healthchecks to `wget` probes for runtime alignment. Updated compose healthchecks to `wget` probes for runtime alignment. Source: llm_adapter@2026-05-21 Confidence: low	—

Full changelog

ContextLattice Public v3.3.25

Runtime updates

Migrated fastembed-sidecar from Python/Uvicorn to native Rust service.
Migrated fastembed-gate-refresh runner from Python to native Rust service.
Updated compose healthchecks to wget probes so health is runtime-aligned with non-Python sidecars.

A/B benchmark snapshot (same model/cache)

| lane | old cold | new cold | old warm avg | new warm avg | old warm p95 | new warm p95 |
|---|---:|---:|---:|---:|---:|---:|
| fastembed /embed | 5.465659s | 4.600979s | 0.073445s | 0.020961s | 0.166389s | 0.045181s |

Operational result

http://127.0.0.1:8075/status healthy after restart: 9/9 services, strict no-python-runtime unchanged for gateway lanes.

Breaking Changes

Removed Python/Uvicorn implementation of `fastembed-sidecar` and replaced it with a native Rust service.
Removed Python implementation of `fastembed-gate-refresh` runner and replaced it with a native Rust service.

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 sheawinkler/ContextLattice

Get notified when new releases ship.

About sheawinkler/ContextLattice

Private-by-default memory and context layer for agents with Go/Rust runtime, staged retrieval across fused data backends, and long-horizon context continuity.

All releases →