amx-cli 0.12.3

Agentic Metadata Extractor — AI-powered CLI to infer and manage database metadata

Stop staring at transactions.posting NUMBER(8) wondering what it means.

AI-powered guidance and reference for data analysts, data engineers,
and catalog owners working with undocumented database schemas.

Documentation · Quickstart · Changelog · Issues

---

AMX walks your database, reads your documentation and codebase, then drafts a description for every table, view, and column — with confidence scores and a human review before anything lands in the live database. Three independent sub-agents (Profile, RAG, Code) gather evidence, an orchestrator merges and ranks them, you accept / edit / skip, and AMX writes approved descriptions back as native COMMENT statements on the engine.

Five minutes from pip install to your first reviewed description. Ten supported database backends, seven LLM providers.

Install

pip install amx-cli

The PyPI distribution is amx-cli; the import package is amx (import amx). Requires Python 3.10+. See the installation guide for prerequisites, source builds, and where AMX writes config / history / logs.

Quick start

amx                       # open the interactive session (the AMX REPL)
/setup                    # one-time wizard: DB profile + LLM profile
/connect                  # sanity-check the active connection
/run core.transactions          # generate suggestions, review, accept
/apply                    # write approved descriptions back to the database

/run without an argument opens a scope picker (Database / Schema / Asset). /run-apply short-circuits review-and-apply when you already trust the model. If anything misbehaves, amx doctor runs from any shell — even when AMX itself can't start — and prints actionable hints next to each ✗.

The full guided walkthrough is at the 5-minute quickstart and first-run walkthrough.

What you get

Cryptic identifier in:

core.transactions.posting   NUMBER(8) NULL

Reviewed description out:

Posting date. The accounting period this transaction lands in, encoded
as YYYYMMDD. Distinct from the system-level effective date (eff_dt)
that records when the row physically arrived in the warehouse.

  confidence: high · logprob: 0.91 · sources: code (3 refs), docs, db profile

Every column gets up to N ranked alternatives, every suggestion is grounded in evidence (db profile, code references, doc snippets), and every approval is recorded in local run history that you can re-evaluate later with /history review.

Supported database backends

PostgreSQL · Snowflake · Databricks (Unity Catalog) · BigQuery · MySQL / MariaDB · Oracle · SQL Server · Redshift · ClickHouse · DuckDB

Per-backend setup, connection details, and the capability matrix live in the Backends section.

Supported LLM providers

OpenAI · Anthropic · Google Gemini · DeepSeek · OpenRouter · Ollama · vLLM / LM Studio · any OpenAI-compatible endpoint

Provider-specific guides (including OpenAI / Anthropic Batch mode and local-model setups) live in the LLM providers section.

Documentation

Full user, operator, and contributor docs live at omeryasirkucuk.github.io/amx-docs — concepts, the slash-command map, configuration, data sources, collaboration, troubleshooting, and the Python API for headless use. Release notes are in CHANGELOG.md and on the GitHub Releases page.

Contributing & support

• Contributing guide — development setup, branching, commit format, release process
• Security policy — how to report a vulnerability
• Open an issue — bugs, questions, feature requests

License

Apache-2.0 — see LICENSE.