AI-powered Streamlit app that tailors academic CVs to PhD, postdoc, and research vacancies.

Maps AtScale semantic-model elements onto BFO 2020 categories; emits a grounding overlay and RDF for reasoning

One command to stand up an AI-queryable AtScale semantic-layer POC, built on @atscale/ps-utils

Explain why a metric moved: walk the AtScale concept graph for derivation paths whose parts moved the same way

Offline tool that turns ungroundable NL questions into draft SML YAML constructs via concept-graph inference

Filter the AtScale MCP trace store for high-quality MQO interactions, exported as OpenAI chat JSONL for fine-tuning

Streamlit chat driving Claude over the AtScale semantic layer: plain-English question in, governed chart out, no SQL

Easy way to import the HoleFoods Sample into InterSystems IRIS using an AtScale Sample Bundle

Pure Rust guardrail that rejects MQO queries naming unmapped catalog fields before execution, with repair suggestions

Durable append-only JSONL journal for MCP investigation sessions — survives server restarts, supports replay

Keeps query result rows out of an LLM's context: stores them, returns an opaque handle. MemStore default, DuckDB opt-in.

Concurrent TCP health probe for a fleet of AtScale clusters; one JSON report, exit code you can gate on

Authorization gate for semantic-layer agents: route to the PII-safe model variant, or deny by default

Result-set eval harness: scores an NL agent against an AtScale semantic layer on the rows it returns, not the query

Claude Code skill: drafts and reviews enterprise B2B data-platform PRDs — template, artifact tree, migration contract

Merge describe_model from many AtScale clusters into one catalog, with provenance and conflict policies

Typed TOML/JSON schema for routing across N AtScale clusters: endpoint, auth, backends, priority; parse, validate, query

Deterministic CLI that proposes bound-ready follow-up MQOs from session + model hierarchy graph — no LLM, no network

CLI that diffs two AtScale describe_model catalogs and flags semantic changes (expression, aggregation) vs cosmetic ones

Rust verdict engine: does a converted measure's result match its Power BI reference under DAX semantics, not float equality?