meta-authoring
Standardizes dbt metadata to ensure assets are searchable, governed, and correctly routed by Nova.
Install
mkdir -p .claude/skills/meta-authoring && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/17966" && unzip -o skill.zip -d .claude/skills/meta-authoring && rm skill.zipInstalls to .claude/skills/meta-authoring
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Builds and reviews high-signal `meta.nova` and curated domain references for dbt resources. Use when adding Nova metadata, choosing canonical datasets or indicators, defining model-bound measures or metric templates, annotating columns, de-ranking helper models, authoring domain reference docs, or improving Nova MCP search and metadata scores.Key capabilities
- →Author or review `meta.nova` for dbt resources.
- →Classify entities as canonical dataset, helper, metric template, or source.
- →Use entity metadata for stable routing and governance signals.
- →Use column metadata for identifiers, time fields, and high-signal dimensions.
- →Verify referenced fields exist in `meta.nova`.
How it works
The skill guides the authoring and review of `meta.nova` fields by classifying entities, defining business concepts, and applying specific rules for metadata placement and content.
Inputs & outputs
When to use meta-authoring
- →Add nova metadata to dbt models
- →Annotate dbt columns
- →Define canonical datasets
- →Review metadata consistency
About this skill
Meta Authoring Skill (dbt-nova)
Mission
Author or review meta.nova so Nova can route agents to the right dbt resource, business definition, grain, and governance signal without adding noisy metadata.
Good metadata is:
- structurally valid
- stable business intent, not one-off report logic
- attached to the execution entity that owns the data or KPI
- discoverable through Nova search
- minimal enough to stay maintainable
Required Frame
Before editing, state:
- entity classification: canonical dataset, helper/staging/intermediate, metric template, source, or column annotation
- business concept: dataset, measure, metric, dimension, identifier, time field, or governance signal
- existing surface: current indicators, columns, overlapping entities, and canonical definitions
- intended surface: entity fields,
measures,metric/metrics, columnmeta.nova, orsearch.candidates - validation path: local schema/audit checks if editing files, then Nova search/score checks after manifest refresh
Do not author a new canonical measure, metric, or dataset until the repeated-concept surface has been checked.
Tool Strategy
Prefer Nova MCP for evidence:
show_metadatato confirm the loaded project and manifest timestampsearch_indicatorandindicator_inventoryfor measures and metricssearch_columnsandcolumn_inventoryfor column semanticssearch,get_entity,get_context, andget_columnsfor current contractscompare_grains,find_entity_overlap, ormodelling_consistency_reportwhen placement is unclearget_metadata_scoreandget_undocumentedfor review and scoring evidence
Use the CLI only when you are validating local file edits or a local manifest build. A hosted MCP endpoint reflects its deployed manifest; it cannot prove uncompiled local YAML changes and should not be reloaded from this skill.
Authoring Rules
- Use entity metadata for stable routing:
canonical,domains,use_cases,synonyms,tier,grain, andgovernance. - Use
measureswhen an aggregation belongs on the execution model. - Use
metricormetricsfor reusable KPI templates; never set both on the same entity. - Use column metadata only for identifiers, time fields, high-signal dimensions, measure columns, and genuine semantic disambiguation.
- Use
search.candidates.analyst: falseonly for helper, ops, staging, or intermediate resources that should remain searchable but not rank first for analysts. - Use canonical flags sparingly. Repeated business terms should have one preferred definition at the relevant level.
- Keep synonyms concise: 2-8 business terms is usually enough.
- Verify referenced fields exist:
grain.time_field,grain.dimensions,measures[].field, andrecommended_filters[].field. - Treat PII and compliance as first-class governance signals. Prefer entity-level
governancefor dataset sensitivity and column-level governance only where a specific exposed column needs it. - Use curated domain references for narrative guidance that does not belong in
meta.nova: business context, scope and exclusions, standard hygiene filters, canonical-entity rationale, gotchas, and cross-domain handoffs. Domain references must point back to Nova metadata rather than replacing it. - Do not use raw query corpora, notebook archives, or dashboard history as the source of authority for domain references. They can be supporting evidence only after the governed metadata and eval evidence are clear.
Workflow
- Read
references/workflow.md. - Read one transport reference:
references/transport-mcp.mdfor live Nova evidence, orreferences/transport-cli.mdfor local validation. - Load
references/decision-rules.mdwhen choosing a Nova surface or canonical placement. - Load
references/patterns.mdonly when writing YAML. - Load
references/domain-reference-template.mdwhen authoring or reviewing a domain reference document. Usereferences/domain-reference-example.mdonly when a concrete synthetic example is useful. - Load
references/review-checklist.mdbefore final handoff.
Output Standard
Final handoffs must include:
- classification and intended Nova surface
- metadata added or changed
- domain reference sections added or changed, if any
- repeated-concept and canonicality evidence
- validation result
- search or score verification result
- remaining caveats or follow-up cleanup
Do not dump full YAML unless the user asks for it or the review requires exact patch text.
When not to use it
- →When the task does not involve adding or reviewing Nova metadata or domain references.
- →When the user wants to use raw query corpora, notebook archives, or dashboard history as the source of authority for domain references.
- →When the user wants to dump full YAML without explicit request or review necessity.
Limitations
- →The skill focuses on `meta.nova` and curated domain references for dbt resources.
- →The skill requires structural validity and stable business intent for metadata.
- →The skill uses Nova MCP for evidence gathering.
How it compares
This skill provides a structured approach to authoring high-signal `meta.nova` and curated domain references, ensuring discoverability and maintainability, unlike ad-hoc metadata additions.
Compared to similar skills
meta-authoring side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| meta-authoring (this skill) | 0 | 20d | No flags | Advanced |
| database-documentation-gen | 2 | 10d | Review | Intermediate |
| project-logger | 0 | 4mo | Review | Beginner |
| domain-model-documentation | 0 | 4mo | No flags | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by joe-broadhead
View all by joe-broadhead →You might also like
database-documentation-gen
jeremylongshore
Process use when you need to work with database documentation. This skill provides automated documentation generation with comprehensive guidance and automation. Trigger with phrases like "generate docs", "document schema", or "create database documentation".
project-logger
diegosouzapw
SQLite-based project documentation logger for tracking API references, components, and project progress. Use this skill when documenting code changes, adding API documentation, recording component updates, or tracking project milestones. Automatically invoked when user mentions documentation, change
domain-model-documentation
DABO-ERP
Use this skill when building or reviewing the domain model for the hotel/hostel management system. This includes defining entities, attributes, relationships, and enumerations based on hospitality domain expertise.
add-neon-docs
neondatabase-labs
Use this skill when the user asks to add documentation, add docs, add references, or install documentation about Neon. Adds Neon best practices reference links to project AI documentation (CLAUDE.md, AGENTS.md, or Cursor rules). Does not install packages or modify code.
sql-optimization-patterns
wshobson
Master SQL query optimization, indexing strategies, and EXPLAIN analysis to dramatically improve database performance and eliminate slow queries. Use when debugging slow queries, designing database schemas, or optimizing application performance.
drizzle-orm
EpicenterHQ
Drizzle ORM patterns for type branding and custom types. Use when working with Drizzle column definitions, branded types, or custom type conversions.