diff --git a/README.md b/README.md index acee041..90fe960 100644 --- a/README.md +++ b/README.md @@ -22,7 +22,7 @@ Extras: `labagent[rag]` (Databricks RAG), `[rag-local]` (laptop RAG), `[rag-lab] |---|---|---| | `azure-labs` | Copy Azure VMs, disks and network from one resource group to another with snapshots, or snapshot a dead VM or disk. Source is read-only. An Azure Lab Services alternative. [Guide](docs/azure-labs.md) | Azure (needs `az login`) | | `rag-sec-tutorial` | Nine notebooks: RAG on SEC 10-K filings with embeddings, pgvector, hybrid search, citations, an evaluation gate, a deployed app | Databricks Free Edition | -| `db-agent` | [db-agent](https://github.com/db-agent/db-agent) text-to-SQL app on Databricks Apps, pinned commit | Databricks Free Edition | +| `db-agent` | [db-agent](https://github.com/db-agent/db-agent) text-to-SQL app on Databricks Apps, pinned commit. [Which model, warehouse and schema it picks](docs/packs.md) | Databricks Free Edition | | `dbx-lakehouse` | Medallion lakehouse (bronze, silver, gold) from real open data, with masks, a row filter, notebooks and exercises. [Guide](docs/dbx-lakehouse.md) | Databricks Free Edition, AWS, Azure | | `dbx-workspace` | Premium Databricks workspace via Terraform. Plan only unless `apply=yes`. [Guide](docs/dbx-workspace.md) | AWS or Azure (costs money) | | `aml` | Synthetic anti-money-laundering lab with an answer key: RAG for developers, Microsoft 365 Copilot for staff. [Guide](docs/aml.md) | Your machine | diff --git a/docs/packs.md b/docs/packs.md index a17c05c..e4325b6 100644 --- a/docs/packs.md +++ b/docs/packs.md @@ -87,7 +87,11 @@ labagent destroy db-agent --name demo Steps: `preflight` (login, SQL warehouse, sample data, serving endpoints), `fetch` (pinned commit), `token`, `app`, `verify` (asks a real question through the deployed app: this proves the token, warehouse, LLM and SQL safety layer work together). -- **Pinned code.** `ref` must be a full commit SHA, and the clone is refused if HEAD differs. To move to a newer db-agent, review it and change the pin. +- **What it picks.** Chat model: `--param chat_endpoint`, else the first serving endpoint named like `llama|gpt|claude|gemma|qwen|instruct|chat`. Embedding model: + `--param embedding_endpoint`, else the first named like `gte|bge|embed`. The app calls `/serving-endpoints` with its token. Data: `catalog` (default `samples`) + and `dataset` (default `bakehouse`); preflight fails with the list of schemas if it is missing. Warehouse: `--param warehouse` (id or name), else the first one. + These are first-match guesses: pass the names yourself when the workspace has several. Preflight records the choices as `chat_endpoint`, `embedding_endpoint` and `warehouse_id`. +- **Pinned code.** `ref` must be a full commit SHA, and the clone is refused if HEAD differs. To move to a newer db-agent, review it and change the pin (`PINNED` in `labagent/packs/db_agent/pack.py`), or try one commit with `--param ref=`. - **Data.** The `samples` catalog already exists in every workspace, so nothing is downloaded or redistributed. - **Credential.** The app needs a token for the warehouse and the LLM. labagent creates one with a 30-day lifetime (`--param token_days`), stores it only as an app secret (via stdin, never on a command line, in state or in output), and `destroy` revokes it. If your workspace blocks diff --git a/labagent/packs/db_agent/pack.py b/labagent/packs/db_agent/pack.py index b342e68..8ee7e37 100644 --- a/labagent/packs/db_agent/pack.py +++ b/labagent/packs/db_agent/pack.py @@ -26,7 +26,7 @@ from ..databricks.models import CHAT_RE, EMBED_RE REPO = "https://github.com/db-agent/db-agent" -PINNED = "eef0a308d052b6bdd00fdfec7ee1c9a61c01426f" # reviewed commit; change it deliberately (see docs/packs.md) +PINNED = "15150ab87b2d50ed8fee71408afbd2f6aba54a14" # reviewed commit; change it deliberately (see docs/packs.md) SECRET_KEY = "token" SMOKE_QUESTION = "How many customers are there?" IGNORE = shutil.ignore_patterns("node_modules", "dist", ".env", ".env.*", "knowledge.json", "*.log", "__pycache__")