Google Cloud Data Lineage Summary
Summarize BigQuery and GCS lineage to debug data quality and provenance.
The skill is explicitly scoped to read-only lineage queries and summaries, and discloses its Google Cloud Data Lineage MCP dependency, authentication configuration, and query boundaries. No malicious behavior or write operation is shown. However, querying every physical region may broaden data visibility, while least-privilege requirements, user confirmation, sensitive-data handling, credential isolation, and recovery controls are unspecified, so points are deducted.
The workflow requires separate upstream and downstream calls and gives concrete parameters, field-level options, and output structure; it also includes MCP configuration guidance. Static materials contain no tests, CI evidence, abnormal-input handling, pagination/limit strategy, or diagnosable failure feedback, so the score is conservative and reduced.
The name, intended use, and non-use cases are reasonably clear, covering table- and column-level lineage while distinguishing generic queries and impact analysis. Trigger rules, input validation, permission/region failure behavior, Chinese-language output, and mainland-China network reachability are not established, so points are deducted.
The documentation includes metadata, prerequisites, staged workflow, parameter examples, external references, installation entry points, an Apache-2.0 license, and an issue-reporting path. It lacks skill-specific versioning, changelog, named maintenance responsibility, compatibility notes, FAQs, and systematic troubleshooting, so points are deducted.
The intended Markdown report is directly readable and specifies upstream/downstream content, asset naming, and query metadata, so the core task is well defined. No real output, third-party execution evidence, or comparison with manual alternatives is supplied; high-limit all-region queries may also create cost and output-size issues, so the score remains below the static ceiling.
The materials provide the MCP tool name, request examples, configuration fragment, and official documentation references, giving the core flow some auditability. There is no committed test suite, CI result, fixed fixture, or independent reproduction record, so evidence coverage is limited and the score is low.
- Operation depends on the Google Cloud Data Lineage MCP, Google credentials, and a reachable remote service; verify permission scope, regional reachability, and credential isolation before deployment.
- All-region queries with a high maxResults limit may broaden data exposure, increase latency or cost, and produce oversized results; consider scoped locations, permissions, and truncation handling.
- Behavior is not specified for MCP unavailability, insufficient permissions, invalid regions, truncated results, or retryable failures.
What does this skill do, and when should you use it?
This skill analyzes Google Cloud Data Lineage for BigQuery and Google Cloud Storage assets. It queries both upstream and downstream lineage from a focal table or field, then turns the graph into an easy-to-read Markdown walkthrough. It requires the Google Cloud Data Lineage (Knowledge Catalog) MCP Server and dynamically retrieves supported physical regions. It fits provenance and data-quality investigations, but not lineage editing, downstream deprecation, or blast-radius analysis.
For a specified table or field, it uses the Google Cloud Data Lineage MCP Server to make separate UPSTREAM and DOWNSTREAM search_lineage calls. Before querying, it uses read_url to retrieve the current Knowledge Catalog physical-region list and applies defaults of maxDepth=10, maxResults=5000, and maxProcessPerLink=10. It supports table-level lineage with column-level links through a wildcard field and supports focused column-level queries. It returns a Markdown summary covering the inferred flow, systems, upstream and downstream paths, and query metadata.
- A data engineer tracing a BigQuery table’s sources while investigating a data-quality issue.
- An analyst trying to understand upstream inputs and downstream consumers for a BigQuery or GCS asset.
- A data-governance practitioner reviewing the provenance of a specific column.
- A team converting a complex lineage graph into a clear left-to-right explanation for stakeholders.
What are this skill's strengths and limitations?
- Covers both upstream and downstream flows.
- Supports table-level and column-level lineage summaries.
- Reports locations, parent location, depth, and processing limits to make query boundaries transparent.
- Requires specific asset names when the result contains fewer than five assets, improving verifiability.
- Requires the Google Cloud Data Lineage (Knowledge Catalog) MCP Server.
- Cross-region coverage depends on dynamically retrieving and supplying the supported location list.
- Does not edit lineage, handle downstream deprecation, or perform blast-radius impact analysis.
- The supplied source does not document a test suite or detailed standalone installation and authentication procedures.
How do you install this skill?
Run npx skills add google/skills and select skills/cloud/datalineage-summary during installation. The supplied source documents repository-level installation only; it does not provide a standalone installation command or version requirements for this skill.
How do you use this skill?
After installation, try: "Summarize the upstream and downstream lineage for bigquery:project.dataset.table, including column-level lineage." The skill requires access to a Google Cloud Data Lineage (Knowledge Catalog) MCP Server that can run search_lineage; the source does not document client-specific trigger syntax or authentication steps.
How does this skill compare with similar options?
For downstream blast-radius or asset-impact analysis, the source explicitly points to datalineage-bigquery-asset-impact-analysis; this skill is focused on summarizing lineage graphs.