using-datapowers
Start here. Introduction to the skills library for data mining and statistical rigor.
Using Datapowers
The Workflow
Datapowers enforces a disciplined analytical workflow:
- brainstorming → Hypothesis-first design.
- writing-analysis-plans → Task decomposition.
- data-profiling → Context injection without raw data.
- executing-plans → Two-stage review for every task.
- leakage-guard → Temporal and preprocessing audit.
- test-driven-data-science → Three-layer assertions.
- verification-before-delivery → Evidence-based completion.
Triggering Skills
| Skill | Trigger Keywords |
|---|---|
brainstorming | "Analyze", "Hypothesis", "Design spec" |
analysis-manifest | Session start, "Where are we?", "Resume analysis", after brainstorming completes |
data-profiling | "What's in the data?", "Profile", "New dataset" |
leakage-guard | "Feature window", "Split", "Leakage" |
test-driven-data-science | "Before training", "Assertion", "Drift" |
executing-plans | "Start tasks", "Follow plan" |
requesting-statistical-review | "Audit results", "Significance" |
verification-before-delivery | "Done", "Fixed", "Complete" |
writing-data-skills | "New skill", "Add skill", "Contribute skill" |
The Philosophy
- Statistical Integrity > Model Metrics
- Validation Before Implementation
- Evidence over Assertions
- Isolated Subagents with High-Density Profiles
Ready? Invoke brainstorming to begin your analysis.