Config matrix

The exhaustive, always-current reference for GoldenPipe configuration: the PipelineConfig / StageSpec schema, the status/column/fixer vocabularies, and every GOLDENPIPE_* runtime knob. Built for humans and AI agents.

Everything below the line is generated from code (scripts/gen_config_matrix.py) and verified in CI. The guidance above the line is hand-authored. See Overview for a walkthrough.

How GoldenPipe is configured

GoldenPipe is the orchestrator: a PipelineConfig is a thin envelope over a list of stages, and each stage delegates its real tuning to the suite tool it adapts.

stages — a list of StageSpec{name, use, needs, skip_if, on_error, config}. use names the adapter (e.g. goldencheck.scan, goldenflow.transform, goldenmatch.dedupe, goldenanalysis.report); config is a free-form dict passed straight to that tool. So the deep knobs for a stage live in that tool’s own config matrix — see GoldenMatch, GoldenCheck, GoldenFlow, GoldenAnalysis.
on_error per stage is continue (default) or abort.
Adapter/stage names are packaging entry points discovered at runtime (StageRegistry), so the set is open/extensible and not gated here; the built-ins are load, infer_schema, goldencheck.scan, goldenflow.transform, goldenmatch.dedupe, goldenmatch.dedupe_fused, goldenmatch.identity_resolve, goldenanalysis.report.

Config object reference

Every config object in the pydantic tree(s), generated from the package schema. Nested objects link by name.

`PipelineConfig`

Top-level pipeline configuration.

Field	Type	Default	Choices / notes
`pipeline`	str	required	Name of the pipeline, identifying this end-to-end configuration in runs and logs.
`source`	str \| None	`None`	Optional default input location that stages read from when they do not specify their own source.
`output`	str \| None	`None`	Optional default destination where the pipeline writes its final results.
`stages`	list[StageSpec \| str]	required	`StageSpec` — Ordered list of stages to run, each given as a full StageSpec or a shorthand string naming the adapter to use.
`decisions`	list[str]	`[]`	Recorded pipeline decision entries that capture the choices and rationale behind this configuration.

`StageSpec`

Configuration for a single pipeline stage.

Field	Type	Default	Choices / notes
`name`	str \| None	`None`	Optional human-readable label for this stage, used in logs and as the key other stages reference in their needs list.
`use`	str	required	Identifier of the adapter that runs this stage, such as goldenmatch.dedupe, naming which suite tool to invoke.
`needs`	list[str]	`[]`	Names of stages that must finish before this one runs, defining the dependency order in the pipeline graph.
`skip_if`	str \| None	`None`	Condition expression that, when it evaluates true at runtime, causes this stage to be skipped rather than executed.
`on_error`	Literal	`'continue'`	`continue`, `abort` — What to do when this stage raises an error: continue on to later stages or abort the whole pipeline.
`config`	dict[str, Any]	`{}`	Free-form settings dictionary passed straight through to the named adapter to configure how the tool runs.

CLI

Every command and its options/arguments, generated from the Typer app. choice-typed options list their allowed values.

Command	Option	Type	Default	Notes
`agent-serve`	`--port`	integer	`8250`	Port for A2A server
`init`	`--dir`	text	`'.'`	Directory to create config in
`interactive`	`source`	text	`None`	Optional data file to load (press ‘r’ to run).
`interactive`	`--config`	text	`None`	Optional pipeline YAML config.
`mcp-serve`	`--transport`	text	`'stdio'`	Transport type: stdio or http
`mcp-serve`	`--host`	text	`'0.0.0.0'`	Host for HTTP transport
`mcp-serve`	`--port`	integer	`8250`	Port for HTTP transport
`run`	`source`	text	required	Input file path
`run`	`--config`	text	`None`	Pipeline YAML config
`run`	`--output`	text	`None`	Output file path
`run`	`--verbose`	boolean	`False`	Show reasoning and timing
`run`	`--identity-path`	text	`None`	Enable Identity Graph; persist to this SQLite path
`run`	`--identity-dataset`	text	`None`	Identity Graph dataset namespace
`run`	`--identity-source-pk`	text	`None`	Column to derive `{source}:{source_pk}` record_id from
`run`	`--identity-weak-threshold`	float	`None`	Auto-flag bottleneck pair as conflicts_with when cluster confidence < this. Defaults to 0.6 (per IdentityConfig).
`serve`	`--port`	integer	`8000`	Port for REST API
`validate`	`--config`	text	required	Pipeline YAML config

MCP tools

4 MCP tool(s) exposed by goldenpipe.mcp.server — the programmatic / agent surface. Config-bearing tools take the same knobs as above.

Tool	Description
`explain_pipeline`	Explain what a pipeline config does
`list_stages`	List all discovered pipeline stages
`run_pipeline`	Run a pipeline on a file OR inline data (records / csv_text) and return per-stage status, reasoning and timing PLUS the deduped output (golden-record count + preview, unique/duplicate counts, cluster count, match stats). Give a `stages` list or `config_path` to control the chain, or omit both for zero-config auto.
`validate_pipeline`	Validate pipeline wiring

Enumerated vocabularies

Allowed values for the str-typed / registry-backed fields above.

Stage status

StageStatus — per-stage result status.

Value	Meaning
`failed`	Stage raised an error.
`skipped`	Stage skipped by its skip_if condition.
`success`	Stage completed successfully.

Pipeline status

PipeStatus — run result status.

Value	Meaning
`failed`	One or more stages failed the run.
`partial`	Some stages succeeded and some failed.
`success`	Every stage succeeded.

Column types

ColumnType — column classification.

Value	Meaning
`address`	Postal address.
`date`	Date/time value.
`description`	Long free text.
`email`	Email address.
`geo`	Geographic coordinate.
`identifier`	Unique identifier / key.
`name`	Person or entity name.
`numeric`	Numeric measure.
`phone`	Phone number.
`string`	Free-form string.
`zip`	Postal/ZIP code.

Cardinality bands

CardinalityBand — column cardinality band.

Value	Meaning
“	Unset / unknown.
`high`	Many distinct values (near-unique).
`low`	Few distinct values.
`mid`	Moderate distinct values.
`skip`	Excluded from cardinality analysis.

Repair fixers

FIXERS — repair-op allowlist.

Value	Meaning
`date_parse`	Parse a date into ISO form.
`email_canonical`	Canonicalize an email (dedupe dots/plus).
`email_normalize`	Lowercase/normalize an email.
`fix_mojibake`	Repair UTF-8/Latin-1 mojibake.
`name_proper`	Proper-case a name.
`normalize_unicode`	Normalize to Unicode NFC.
`phone_national`	Normalize a phone to national format.
`zip_normalize`	Normalize a ZIP code.

Stages

BUILTIN_STAGES — StageSpec.use.

Value	Meaning
`goldenanalysis.report`	Run GoldenAnalysis metrics + reporting.
`goldencheck.scan`	Run a GoldenCheck data-quality scan.
`goldenflow.transform`	Run GoldenFlow transforms / standardization.
`goldenmatch.dedupe`	Run GoldenMatch dedupe / entity resolution.
`goldenmatch.dedupe_fused`	Run the fused GoldenMatch dedupe kernel.
`goldenmatch.identity_resolve`	Resolve records against the identity graph.
`infer_schema`	Infer the source schema (via infermap).
`load`	Load a source file into the pipeline frame.

Environment variables (index)

1 GOLDENPIPE_* runtime knob(s) read by the package, scanned from source so this is complete. Grouped by area:

NATIVE (1): GOLDENPIPE_NATIVE

Get started

Concepts

GoldenMatch

GoldenCheck

GoldenFlow

GoldenPipe

GoldenAnalysis

InferMap

SQL extensions

Reference

Research

How GoldenPipe is configured

Config object reference

`PipelineConfig`

`StageSpec`

CLI

MCP tools

Enumerated vocabularies

Stage status

Pipeline status

Column types

Cardinality bands

Repair fixers

Stages

Environment variables (index)

See also

​How GoldenPipe is configured

​Config object reference

​PipelineConfig

​StageSpec

​CLI

​MCP tools

​Enumerated vocabularies

​Stage status

​Pipeline status

​Column types

​Cardinality bands

​Repair fixers

​Stages

​Environment variables (index)

​See also

How GoldenPipe is configured

Config object reference

`PipelineConfig`

`StageSpec`

CLI

MCP tools

Enumerated vocabularies

Stage status

Pipeline status

Column types

Cardinality bands

Repair fixers

Stages

Environment variables (index)

See also