SLayer, a semantic layer maintained by your agent

v0.7.1 Breaking

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

Published 1mo 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

semantic-layer

ReleasePort's take

Light signal

editorial:auto 1mo

SLayer v0.7.1 adds SQL Server support and an openCypher filter while breaking the `search()` response format.

Why it matters: The `search()` API now returns a single flat results list, removing per‑bucket caps (severity 70).

Summary

AI summary

SQL Server added as first-class datasource and search API flattened with cypher_filter support.

Changes in this release

Type	Severity	Summary	CVE
Breaking	High	Changes `search()` to return a single flat `results` list, removing per‑bucket caps. Changes `search()` to return a single flat `results` list, removing per‑bucket caps. Source: llm_adapter@2026-06-08 Confidence: low	—
Breaking	Medium	Changes `search()` response to a single flat `results` list capped by `max_results`, removing separate `memories`, `example_queries`, and `entities` buckets. Changes `search()` response to a single flat `results` list capped by `max_results`, removing separate `memories`, `example_queries`, and `entities` buckets. Source: granite4.1:30b@2026-06-08-audit Confidence: low	—
Feature
Feature	Medium	Adds SQL Server (T‑SQL) as a first‑class datasource with auto‑generated date math and type mappings. Adds SQL Server (T‑SQL) as a first‑class datasource with auto‑generated date math and type mappings. Source: llm_adapter@2026-06-08 Confidence: high	—
Feature	Medium	Adds `cypher_filter` parameter to `search()` for openCypher graph narrowing. Adds `cypher_filter` parameter to `search()` for openCypher graph narrowing. Source: llm_adapter@2026-06-08 Confidence: high	—
Feature	Low	Increases column hit sample values in `search()` results from ~20 to top‑50 and adds distinct count line. Increases column hit sample values in `search()` results from ~20 to top‑50 and adds distinct count line. Source: llm_adapter@2026-06-08 Confidence: high	—
Dependency	Low	Renames pip extra from `embedding_search` to `advanced_search`, now includes graph backend. Renames pip extra from `embedding_search` to `advanced_search`, now includes graph backend. Source: llm_adapter@2026-06-08 Confidence: high	—
Bugfix	Medium	Corrects parenthesization of filters referencing multi‑term `Column.sql` expressions. Corrects parenthesization of filters referencing multi‑term `Column.sql` expressions. Source: llm_adapter@2026-06-08 Confidence: high	—
Bugfix	Medium	Fixes PostgreSQL error when ordering by a `ModelMeasure` name in query. Fixes PostgreSQL error when ordering by a `ModelMeasure` name in query. Source: llm_adapter@2026-06-08 Confidence: low	—

Full changelog

SQL Server (T-SQL) is now a first-class datasource: configure with type: mssql (or sqlserver / tsql), and SLayer auto-generates T-SQL flavoured date math, stat aggregations, and full type mappings. A Docker Compose example with SQL Server 2022 ships under examples/sqlserver/.
search() accepts a new cypher_filter parameter - an openCypher MATCH ... RETURN n.id AS id query that narrows results to a subset of your semantic graph (e.g. "only columns reachable from the orders model") before any ranking happens.
search() now returns a single flat results list capped by max_results, instead of three separate memories/example_queries/entities buckets with per-bucket caps. If you were reading the bucket fields or passing max_memories / max_example_queries / max_entities, update your call sites - you can use the Cypher filter to narrow down the search to a particular set of entities.
The pip extra for embedding-based search has been renamed from embedding_search to advanced_search (it now also pulls in the graph backend for cypher_filter). Install with pip install motley-slayer[advanced_search].
Column hits from search() now surface the full top-50 sample values (was previously truncated to ~20) plus a Distinct count: N line when the underlying column has more distinct values than the snapshot captured.
Fix: ordering a query by a ModelMeasure referenced by name (e.g. order=[{"column": "aov"}]) no longer fails on PostgreSQL with an undefined-column error - the ORDER BY alias now matches the SELECT alias.
Fix: filters that reference a multi-term Column.sql (arithmetic composites, AND/OR expressions, NOT, BETWEEN, IN) now emit correctly parenthesised SQL. Previously, a filter like <col> IS NULL against a column defined as archived AND deleted could parse as archived AND (deleted IS NULL) instead of the intended (archived AND deleted) IS NULL.

Breaking Changes

`search()` now returns a single flat `results` list; previous separate `memories`, `example_queries`, and `entities` buckets are removed. Update call sites accordingly.
The pip extra for embedding-based search renamed from `embedding_search` to `advanced_search`. Install with `pip install motley-slayer[advanced_search]`.

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 SLayer, a semantic layer maintained by your agent

Get notified when new releases ship.

About SLayer, a semantic layer maintained by your agent

All releases →

Related context

Related tools

Earlier breaking changes

v0.6.3 Datasource names now reject dots, slashes, nulls, empty/whitespace; existing names containing '.' will fail validation on upgrade.
v0.6.0 recall_memories surface entirely removed with no deprecation shim.
v0.5.1 Two-mode reference semantics enforced: SQL mode accepts arbitrary SQL; DSL mode strictly resolves identifiers.
v0.5.1 RecallHit.match_count renamed to RecallHit.score across MCP, REST, CLI, and SlayerClient.