skills.rest
open-metadata
open-metadataOFFICIAL
Explore all skills by this author →
Website𝕏@open_metadataGitHub

connector-audit

Audits OpenMetadata connectors through seven investigation prompts covering metadata, errors, auth, lineage, and scale.

15.1k|2.3k|Updated Aug 1, 2021
Search Tags:#audit#data ingestion#connector#Reliability#OpenMetadata#LINEAGE#Code review
One-click install
npx skills add https://github.com/open-metadata/OpenMetadata --skill connector-audit
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: connector-audit
Source: https://github.com/open-metadata/OpenMetadata/tree/main/skills/connector-audit
Command: npx skills add https://github.com/open-metadata/OpenMetadata --skill connector-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Evaluating whether an OpenMetadata connector meets reliability standards requires deep, manual code review across metadata ingestion, error handling, authentication, lineage, and performance — a process that is slow, inconsistent, and easy to do superficially.

Core Features & Use Cases

  • Seven-Prompt Investigation Workflow: Runs setup, five parallel investigation prompts (metadata ingestion, error handling, connection/auth, lineage, scale/performance), a synthesis prompt, and an implementation prompt against connector standards.
  • Structured Audit Reports: Saves per-prompt findings with file:line references, severity ratings, and rating calibration to .claude/audit-results/ for cross-validation and planning.
  • Fix Planning and Implementation: Consolidates findings into a prioritized refactor plan with PR scoping, then implements fixes with tests, or produces a dry-run plan with before/after diffs.
  • Use Case: Before making major changes to the MySQL connector, run a full audit to identify missing pagination, broken column-level lineage, and absent token refresh, then get a sequenced PR plan to fix them.

Quick Start

Ask the AI to run a connector reliability audit on the mysql connector and follow the guided seven-prompt workflow.

Frequently Asked Questions about connector-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit an OpenMetadata connector for reliability?▼

Invoke the audit with a connector name such as mysql or snowflake to run the full seven-prompt workflow. It investigates metadata ingestion, error handling, auth, lineage, and scale, then synthesizes findings into a prioritized implementation plan.

What is the difference between connector-audit and connector-review?▼

connector-audit performs a deep seven-prompt reliability investigation over hours, covering the full connector plus base classes. connector-review is a breadth check for PRs using parallel agents over changed files only, completing in minutes.

Can I run only one audit prompt instead of the full audit?▼

Yes, use the --prompt N flag to run a single prompt from 1 to 7, --prompts N,M for specific prompts, or --from N to run prompts N through 7. This is useful for re-running an area after fixes.

Does the connector audit modify connector source code?▼

Only in the final implementation prompt (P7), which writes code, runs tests, and creates commits based on the approved plan. With the --dry-run flag, P7 produces before/after diffs and test code without writing any code.

Where are connector audit results saved?▼

Reports are saved as Markdown files under .claude/audit-results/, one per prompt (01 through 07), with connector context stored in .claude/connector-audit.json. Each report requires user approval before being saved.

Related Skills

Explore More Software Engineering Skills→
mako-aimako-ai
11

add-connector

Scaffold a connector module with standardized structure, tests, and CDC/webhook support.

Official
Advanced
luicabref97luicabref97

tdd:test-driven-development

Enforce a red-green-refactor cycle with failing tests before production code.

Community
Intermediate
Obsidian-OwlObsidian-Owl
4

speckit-test-review

Assess test quality across unit, integration, contract, and e2e tests.

Official
Advanced
danchamorrodanchamorro
13

code-review

Review code changes and pull request diffs for bugs, security vulnerabilities, and quality issues.

Community
Advanced