feat: add workflow templates (#9)

9 bundled workflow templates:
- data-pipeline: ETL extract/transform/validate
- content-review: multi-agent draft/quality/accuracy/edit
- customer-support: tiered triage/response/escalation
- code-review: parallel correctness+security+style analysis
- research: research/fact-check/summarize
- content-pipeline: research → write → edit article creation
- translate-polish: auto-detect language, translate, review
- brainstorm: ideation → evaluation → action plan
- weekly-report: raw notes → organized → polished report

workflows/data-pipeline.toml

@@ -0,0 +1,32 @@
+id = "data-pipeline"
+name = "Data Pipeline"
+description = "ETL pipeline that extracts data from a source, transforms it into a target format, and validates the output for completeness and correctness."
+category = "data"
+tags = ["etl", "data", "pipeline", "transform"]
+
+[[parameters]]
+name = "data_source"
+description = "URL or file path to the data source"
+param_type = "string"
+required = true
+
+[[parameters]]
+name = "output_format"
+description = "Desired output format (json, csv, markdown)"
+param_type = "string"
+required = false
+default = "json"
+
+[[steps]]
+name = "extract"
+prompt_template = "Extract raw data from the following source: {{data_source}}. Return the complete dataset without any transformation, preserving the original structure."
+
+[[steps]]
+name = "transform"
+prompt_template = "Transform the following raw data into well-structured {{output_format}} format. Apply standard cleaning: trim whitespace, normalise dates to ISO-8601, remove duplicate rows, and convert empty strings to null.\n\nRaw data:\n{{extract}}"
+depends_on = ["extract"]
+
+[[steps]]
+name = "validate"
+prompt_template = "Validate the transformed data below for completeness and correctness. Check for: missing required fields, type mismatches, invalid date formats, out-of-range values, and referential integrity. Return a validation report with pass/fail status and any issues found.\n\nTransformed data:\n{{transform}}"
+depends_on = ["transform"]