DataSentry v1.0.4 — Fix official MCP Registry publishing

DataSentry v1.0.4 — Fix official MCP Registry publishing

DataSentry v1.0.4 is a distribution metadata patch that fixes publication to the official MCP Registry.

What changed

• Corrected the canonical MCP Registry identity to io.github.Jackxiaozhiren/datasentry, matching the exact case-sensitive GitHub namespace authorized through OIDC.
• Updated the packaged PyPI ownership marker to the same canonical Registry identity.
• Released the matching package pair:
  • datasentry-ai==1.0.4
  • datasentry-core==0.8.3
• Kept server.json, PyPI package metadata, and the uvx MCP launch path version-aligned.
• Improved the release workflow so permanent Registry errors such as HTTP 400/401/403/409/422 fail immediately instead of being incorrectly treated as PyPI propagation delays.
• Added regression protection and documentation for the case-sensitive Registry namespace requirement.

Scope

This release does not change DataSentry's detectors, repair engine, CLI/API behavior, or data-processing logic. It is a release and MCP distribution correction.

Install

pip install --upgrade datasentry-ai
datasentry demo

MCP

datasentry mcp --project /path/to/project

The canonical official MCP Registry identity is:

io.github.Jackxiaozhiren/datasentry

Validation

The release candidate passed:

• official MCP Registry metadata validation;
• Ruff lint and formatting checks;
• mypy strict;
• full pytest and coverage gate;
• zero-config demo smoke;
• 1M-row benchmark gate;
• CLI, API, and UI smoke tests;
• artifact checks;
• isolated installation of both release wheels.

DataSentry remains focused on the same local-first quality loop:

Find → Explain → Fix safely → Verify

DataSentry v1.0.3 — Official MCP Registry discovery

DataSentry v1.0.3 — Official MCP Registry discovery

DataSentry v1.0.3 makes its local-first data-quality MCP server ready for discovery through the official Model Context Protocol Registry.

What's new

• Added canonical MCP Registry metadata under io.github.jackxiaozhiren/datasentry.
• Added the PyPI ownership marker required for Registry verification.
• Added official mcp-publisher validation to pull requests.
• Added automatic MCP Registry publishing after a successful PyPI release.
• Registry authentication uses GitHub OIDC — no long-lived Registry publishing token is required.
• Added release-time validation to keep the application version, Registry version, PyPI package metadata, and ownership marker aligned.
• Documented the MCP Registry distribution and release workflow.

Install

pip install --upgrade datasentry-ai
datasentry demo

MCP

DataSentry provides a local stdio MCP server:

datasentry mcp --project /path/to/project

Its core safety model remains unchanged:

Find → Explain → Fix safely → Verify

Detection and scoring remain deterministic by default. Repairs are previewed, applied to copies, auditable, reversible, and state-changing MCP operations should remain user-approved.

GitHub Actions quality gate

DataSentry also provides a reusable GitHub Actions quality gate:

jobs:
  datasentry:
    uses: Jackxiaozhiren/datasentry/.github/workflows/datasentry-quality-gate.yml@main
    with:
      path: data/orders.csv
      fail_on: high

Release validation

This release passed the complete DataSentry CI gate, including:

• Ruff lint and formatting
• mypy strict
• pytest and coverage gate
• zero-config demo smoke test
• 1M-row benchmark
• CLI smoke tests
• API and Web UI smoke tests
• artifact validation
• isolated installation of the exact release wheels
• official MCP Registry metadata validation

Full documentation and examples are available in the repository.

v1.0.2 — one-command demo & GitHub onboarding

DataSentry v1.0.2

This release makes DataSentry much easier to understand, try, and integrate into a real data-quality workflow.

Highlights

• New datasentry demo zero-configuration product tour
• Generate dirty data → detect issues → inspect evidence → safely repair → verify
• Runs fully locally — no dataset, API key, cloud service, or LLM required
• Keeps datasentry-demo as a direct console alias
• Reworked onboarding around Find → Explain → Fix safely → Verify
• Scenario-first examples for common workflows
• Copy-paste GitHub Actions data-quality gate
• Stronger MCP onboarding for AI-agent workflows
• Improved package metadata and GitHub project discovery
• Backward-compatible demo artifact behavior

Try it

pip install --upgrade datasentry-ai
datasentry demo

DataSentry keeps deterministic detection as the authority. Repairs are previewed, applied to a copy, verified by re-scanning, and remain reversible.

Why DataSentry

Most data-quality tools work well once you already know which expectations or rules to write.

DataSentry focuses on the step before that too:

Find → Explain → Fix safely → Verify

Discover problems automatically, see the evidence behind them, repair without overwriting the source, and verify that the repair did not introduce regressions.

v1.0.0 — DataSentry 1.0 milestone

Full Changelog: v0.56.0...v1.0.0

v0.56.0 — trends drift parallel view

Full Changelog: v0.55.0...v0.56.0

v0.55.0 — CLI latest consistency

Full Changelog: v0.54.0...v0.55.0

v0.54.0 — rollback failure reasons

Full Changelog: v0.53.0...v0.54.0

v0.53.0 — FIXED deep link + verify loop blog

Full Changelog: v0.52.0...v0.53.0

v0.52.0 — repair diff CLI

Full Changelog: v0.51.0...v0.52.0

v0.51.0 — REST verify + diff endpoints

Full Changelog: v0.50.0...v0.51.0