Expand governed Dataflow editor and node library

README.md

@@ -17,10 +17,40 @@ the previewed row contents.
 - **Risk Compliance:** sanctions matching policy, review, dispositions, and
   legal evidence.
 
-The first implementation supports bounded inline tabular sources and the
-`filter`, `select`, `aggregate`, `sort`, `limit`, and `output` transforms. The
-connector-backed source contract is tracked separately so credentials never
-become part of a pipeline definition.
+## Node Library
+
+The canonical backend catalogue is exposed to the WebUI and groups executable
+nodes by purpose:
+
+| Group | Nodes |
+| --- | --- |
+| Load | Inline data, connector source |
+| Combine | Append rows, join tables |
+| Filter | Filter rows, remove duplicates |
+| Transform | Select columns, derive column, aggregate, sort, limit |
+| Output | Preview output |
+
+Join nodes have explicit left and right ports. Append nodes accept two or more
+inputs. Derived columns use a constrained operation catalogue rather than
+arbitrary code.
+
+Connector sources are resolved through the versioned Core capability
+`connectors.tabular_sources`; Dataflow does not import Connectors or store its
+credentials. Source references pin an expected fingerprint so schema/content
+drift fails visibly instead of silently changing a run.
+
+## SQL And Preview Safety
+
+The SQL workbench parses one `SELECT` statement into the canonical graph. The
+dialect supports projection, aliases, filters, grouping, aggregate functions,
+sorting, limits, `DISTINCT`, and one two-source equi-join. It rejects DDL, DML,
+subqueries, arbitrary functions, file access, and unchecked pass-through
+execution.
+
+Preview reads at most 250 rows per source and enforces time, intermediate-row,
+result-byte, graph-node, and response-row bounds. Saved previews record the
+pipeline revision, executor version, source fingerprints, node diagnostics,
+and output summary, but not source or result rows.
 
 ## Development