Skip to content

Add initial weevr skills: authoring, cli, consumption, medallion#1

Open
brad-cooley wants to merge 2 commits into
mainfrom
feat/initial-skills
Open

Add initial weevr skills: authoring, cli, consumption, medallion#1
brad-cooley wants to merge 2 commits into
mainfrom
feat/initial-skills

Conversation

@brad-cooley

Copy link
Copy Markdown
Contributor

Summary

  • Adds 4 distributable agent skills for weevr project expert building (25 files, ~11,500 lines)
  • weevr-authoring: Thread, weave, loom, and warp YAML authoring with step types reference, config resolution docs, and patterns/examples
  • weevr-cli: CLI commands, configuration, deployment workflows, warp file management, and CI/CD patterns
  • weevr-consumption: Execution modes, Python API, Fabric integration, and troubleshooting
  • weevr-medallion: End-to-end Bronze/Silver/Gold medallion architecture patterns with detailed layer guides

Details

Each skill follows the standard SKILL.md format with YAML frontmatter (name, description), critical notes, table of contents, rules, quick start examples, and a resources/ directory with detailed reference documentation.

Skills are designed for distribution via npx skills add mhs/weevr-skills and integrate with:

  • Fabric MCP Server (@microsoft/fabric-mcp@latest)
  • Fabric REST APIs
  • Existing Fabric skills (spark-authoring-cli, sqldw-, powerbi-, etc.)

Testing

  • YAML examples validated against weevr JSON schemas
  • Skills installable via npx skills add
  • SKILL.md frontmatter parses correctly

Four distributable agent skills for weevr project expert building:
- weevr-authoring: Thread, weave, loom, and warp YAML authoring
- weevr-cli: CLI commands, deployment workflows, CI/CD patterns
- weevr-consumption: Execution modes, Python API, Fabric integration
- weevr-medallion: End-to-end Bronze/Silver/Gold lakehouse patterns

25 files, ~11,500 lines of skill content and reference documentation.
Validated all YAML code blocks across 25 files against the thread, weave,
loom, and warp JSON schemas. Key corrections:

- string_ops: use columns array + expr template, not column-to-operation map
- dimension/fact/naming: nest under target, not at thread top-level
- resolve step: match is an object map, not bare string; use lookup/include/on_unknown
- system_members: sk is integer (not string), code field is required
- fk_sentinel_rate sentinels: use SentinelGroup objects with value field
- connections: add required type: onelake
- lookups: nest source fields under source: object, materialize is boolean
- params not parameters; threads/weaves use array with ref: not map with path:
- pre_steps/post_steps are top-level (no hooks: wrapper)
- quality_gate: min_count not min, expression not custom_sql, no schema_match
- change_detection: named groups with columns/on_change, not type1_attrs/type2_attrs
- warp: remove invalid name/key/enforcement fields
- Various enum corrections (bool not boolean, strategy values, naming conventions)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant