v1.4.1: Data quality and transform tools now exposed via MCP and A2A

[Ben Severn (benzsevern)]

GoldenMatch v1.4.1 adds 3 new MCP tools and 2 new A2A skills so AI agents can inspect and fix data quality before running the matching pipeline.

New MCP Tools (30 total)

Tool	What It Does
scan_quality	Run GoldenCheck scan, return issues (encoding, Unicode, format violations) without fixing
fix_quality	Scan + apply fixes with safe or moderate mode, optionally save output
run_transforms	Run GoldenFlow transforms (phone E.164, dates ISO, categorical spelling, Unicode)

New A2A Skills (10 total)

Skill	What It Does
quality	Scan and fix data quality issues via GoldenCheck
transform	Normalize data formats via GoldenFlow

Why This Matters

Previously, GoldenCheck and GoldenFlow ran silently inside the pipeline. Agents had no way to:

• Run quality checks before committing to a full pipeline run
• Inspect what fixes were applied
• Choose fix aggressiveness (safe vs moderate)
• Run transforms independently

Now they can.

Robustness

All new handlers include:

• File path validation with actionable error messages
• Safe output writes (results preserved if write fails)
• strict=True transform mode (surfaces failures instead of silently returning unmodified data)
• Logging for observability
• Graceful degradation when optional deps not installed

Install: pip install goldenmatch[quality,transform]

PR: #34