From 243f187090f088d2dafa3e70e247d751330b0b88 Mon Sep 17 00:00:00 2001 From: Egor Kraev Date: Wed, 25 Mar 2026 17:06:28 +0100 Subject: [PATCH 1/2] Rewrite README intro with value proposition and roadmap Surface the problem SLayer solves (raw SQL from agents, heavy existing semantic layers) and highlight key differentiators: auto-ingestion, dynamic models, query-time expressions, and time operations. Co-Authored-By: Claude Opus 4.6 --- README.md | 28 +++++++++++++++++++++++++++- 1 file changed, 27 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 088500e..d913056 100644 --- a/README.md +++ b/README.md @@ -5,7 +5,33 @@ [![Docs](https://img.shields.io/badge/docs-readthedocs-blue)](https://agentic-slayer.readthedocs.io/) [![License](https://img.shields.io/github/license/MotleyAI/slayer)](LICENSE) -A lightweight open-source semantic layer for AI agents and humans +A lightweight, open-source semantic layer that lets AI agents query data without writing SQL. + +--- + +### The problem + +When AI agents write raw SQL, things break in production — hallucinated column names, incorrect joins, metrics that drift across queries. Existing semantic layers (Cube, dbt metrics) were built for dashboards: heavy infrastructure, slow model refresh cycles, and limited expressiveness for the kinds of ad-hoc analysis agents need. + +### What SLayer does differently + +- **Auto-ingestion with FK awareness** — Connect a database, and SLayer introspects the schema, detects foreign keys, and generates usable models with denormalized joins instantly. No manual modeling required to get started. +- **Dynamic model manipulation** — Agents create and edit models at runtime. Changes take effect immediately — no rebuild, no deploy, no restart. +- **Query-time expressions** — Compose derived metrics on the fly with the `fields` API (`"revenue / count"`, `"cumsum(revenue)"`, `"change_pct(revenue)"`). No need to pre-define every metric. +- **First-class time operations** — Built-in `time_shift`, `change`, `change_pct`, `cumsum`, `rank`, and `last` — all composable and nestable (e.g., `"change(cumsum(revenue))"`). + +### Roadmap + +- Measures from joined models +- Multistage queries +- Unpivoting +- Smart output formatting (currency, percentages) +- Auto-propagating filters +- Asof joins +- Chart generation (eCharts) +- Claude Code plugin with query skills + +--- ## Quick Start From f8a6abe33c79cc6fdfc51b9b4e10100ed24f5e7a Mon Sep 17 00:00:00 2001 From: Egor Kraev Date: Mon, 30 Mar 2026 11:40:19 +0200 Subject: [PATCH 2/2] Update README.md Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com> --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index d913056..a17fc28 100644 --- a/README.md +++ b/README.md @@ -9,7 +9,7 @@ A lightweight, open-source semantic layer that lets AI agents query data without --- -### The problem +## The problem When AI agents write raw SQL, things break in production — hallucinated column names, incorrect joins, metrics that drift across queries. Existing semantic layers (Cube, dbt metrics) were built for dashboards: heavy infrastructure, slow model refresh cycles, and limited expressiveness for the kinds of ad-hoc analysis agents need.