メニュー
Investigates incorrect, inconsistent, or missing data to identify the corruption source and create a remediation plan. Do NOT use for data migration planning, schema design, or general database optimization.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use to implement a REST or GraphQL API endpoint from a specification or description, including request validation, error handling, and accompanying tests. Activate for building a new API route, resolver, or handler end-to-end. Do NOT use for API design/contract creation, documentation generation, or frontend API consumption.
Use to implement background job processing with a queue, worker, retry with exponential backoff, dead letter queue, and monitoring. Activate when the request involves offloading work from the request cycle (emails, reports, data processing). Do NOT use for real-time event streaming, webhook handling, or scheduled cron jobs.
Use to scaffold a new project from scratch — generating directory structure, configuration files, entry points, and a README skeleton. Activate when starting a new project, service, or package and needing a production-ready starting point. Do NOT use for adding features to existing projects, generating tests, or CI/CD configuration.
Use to design and implement a caching layer for data that is expensive to compute or fetch. Activate for cache-aside, write-through, or write-behind patterns, TTL strategy, and cache invalidation logic. Do NOT use for database query optimization, CDN configuration, or HTTP response caching headers.
Use to generate a CI/CD pipeline configuration for building, testing, and deploying an application. Activate for GitHub Actions, GitLab CI, CircleCI, or similar pipeline tools for any tech stack. Do NOT use for infrastructure provisioning, Dockerfile creation, or deployment strategy design.
Use to generate production-ready code for a new feature, function, class, or module from a description. Activate for writing new code from scratch with error handling, inline documentation, and a self-review checklist. Do NOT use for refactoring existing code, code review, or test generation.
| name | data-corruption-investigation |
| description | Investigates incorrect, inconsistent, or missing data to identify the corruption source and create a remediation plan. Do NOT use for data migration planning, schema design, or general database optimization. |
| version | 1.1.0 |
| time_saved | 30-60 minutes |
| license | Proprietary — Personal Use Only |
| category | debugging |
| complexity | Advanced |
| tokens | ~5500 |
| tags | ["data-corruption","data-integrity","audit-trail","database","debugging"] |
| author | vheins |
Analyze data inconsistencies to identify sources (races, logic errors, partial updates) and produce a remediation and prevention plan.
| Variable | Type | Req | Description |
|---|---|---|---|
tech_stack | string | Yes | DB + App (e.g., "PostgreSQL + Node.js") |
corruption_symptoms | string | Yes | What is wrong, diffs, timeline |
data_samples | string | Yes | Corrupted vs. expected examples |
context | string | Yes | Arch, concurrent writers, audit logs |
| Case | Strategy |
|---|---|
| No Samples | Provide diagnostic queries to identify patterns and scope. |
| External Source | Recommend boundary validation and schema-level data quality checks. |
| Financial Data | High-severity; prioritize reconciliation and involve data engineering. |
flowchart TD
A([Start]) --> B[Parse inputs]
B --> C{Data samples?}
C -- No --> D[Provide SQL queries to identify scope/pattern]
C -- Yes --> F[Analyze Pattern]
D --> E[Enable audit logging]
F --> G[Characterize corruption]
G --> H[Correlate timing]
H --> I{Source?}
I -- Race condition --> J[Concurrent writes]
I -- Missing tx --> K[Partial update]
I -- Logic bug --> L[Calculation error]
I -- External --> M[Third-party import]
I -- Migration --> N[Script bug]
I -- Cascade --> O[Delete/update cascade]
J & K & L & M & N & O --> P[Analyze audit trail]
P --> Q{Financial data?}
Q -- Yes --> R[High-severity/reconciliation approach]
Q -- No --> S[Build Remediation Plan]
R --> S
S --> T[Add Prevention Safeguards]
T --> U([Output: Report])
E --> V([Output: Limited investigation — enable logging & re-run])
@upstash/context7-mcp: Library documentation and examples.@modelcontextprotocol/server-sequential-thinking: Complex reasoning.| Version | Date | Description |
|---|---|---|
| 1.1.0 | 2026-03-20 | Restructured: moved examples/references, added fields |
| 1.0.0 | 2026-03-20 | Initial release |