bi-python-knowledge
Skill2 starsChanged 4 months agoApache-2.0View on GitHub
Python/pandas reasoning and review layer for BI and data agents in the Seshat BI project. Use when an agent must reason about dataframe work — profiling a source, judging dtypes/schema, cleaning and standardizing, merging without fan-out, aggregating at a declared grain, parsing dates, validating/reconciling against SQL and readiness, diagnosing performance/memory, or reviewing a Python pipeline. This is a reasoning layer, not an executor and not a Python tutorial. It does not run pipelines, define metrics, or own gating. See INDEX.md for the live routes.
In this repository60 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- approval-consoleSkill
- approval-evidence-packSkill
- business-knowledge-interviewSkill
- capabilitiesSkill
- consumer-data-dictionarySkill
- cross-table-lineageSkill
- dagster-orchestration-adapterSkill
- dashboard-designSkill
- dashboard-intelligenceSkill
- dbt-transformation-adapterSkill
- evidence-pack-generatorSkill
- first-hour-compassSkill
- friendly-pr-reviewerSkill
- grain-confidence-reviewerSkill
- kpi-contract-builderSkill
- pbi-mcp-write-adapterSkill
- pbip-workflowSkill
- pbip-xraySkill
- pbir-authoring-adapterSkill
- portfolio-watchSkill
- powerbi-dashboard-designSkill
- pr-readiness-reviewerSkill
- readiness-viewerSkill
- release-notes-generatorSkill
- report-intent-interviewSkill
- retail-build-warehouseSkill
- retail-control-roomSkill
- retail-discover-portfolioSkill
- retail-governSkill
- retail-initSkill
- retail-onboard-tableSkill
- retail-orchestrateSkill
- retail-scaffoldSkill
- retail-semantic-checkSkill
- retail-validateSkill
- run-next-readinessSkill
- seshat-studioSkill
- showcase-buildSkill
- source-mappingSkill
- speckit-analyzeSkill
- speckit-checklistSkill
- speckit-clarifySkill
- speckit-constitutionSkill
- speckit-git-commitSkill
- speckit-git-featureSkill
- speckit-git-initializeSkill
- speckit-git-remoteSkill
- speckit-git-validateSkill
- speckit-implementSkill
- speckit-planSkill
- speckit-specifySkill
- speckit-tasksSkill
- speckit-taskstoissuesSkill
- bi-analyst-knowledgeSkill
- bi-bigdata-knowledgeSkill
- bi-dax-knowledgeSkill
- bi-sql-knowledgeSkill
- retail-kpi-knowledgeSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
14 rules. About 0 are common, 11 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈11
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 39
Open
INDEX.md: Do not read the whole knowledge base. - Distinctive: line 44
Stop at boundaries: If the task is metric definition, semantic logic, SQL transformation, or gating, hand off.
- Distinctive: line 62
Execution adapters: run things; they never define meaning, mapping, metrics, semantic logic, or approval.
- Distinctive: line 67
Do not clean data before source meaning and grain are understood.
- Distinctive: line 68
Do not merge dataframes before join keys and cardinality are known.
- Distinctive: line 69
Do not aggregate before declaring grain and additivity.
- Distinctive: line 70
Do not infer business meaning from column names alone.
- Distinctive: line 71
Do not claim reconciliation passed without evidence.
- Distinctive: line 72
Do not fill nulls, coerce dtypes, or join frames without an evidence ledger.
- Distinctive: line 73
Do not write production pipeline code in this layer.
- Distinctive: line 74
Do not use Python to bypass SQL/readiness validation gates.
- Distinctive: line 75
Do not read the whole knowledge base when a router can select files.
- Very specific: line 76
Do not copy book examples or prose.
- Very specific: line 82
All examples use the fictional retail schema in references/retail-dataframe-schema.md.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
BI Python Knowledge Skill
A thin router into a Python/pandas reasoning layer for BI/data agents. It helps an agent think correctly about dataframe work and hand off cleanly into the SQL, DAX, readiness, and dashboard layers. It does not execute code or define business meaning.
Status: core routes live. Dataframe semantics, profiling, dtypes, missing values, cleaning, merge/fan-out, aggregation grain, dates/calendars, validation, performance, active-rule review, positive patterns, and pipeline review all resolve through
INDEX.md.
Use this skill when
Source profiling, dtype/schema issues, cleaning and standardization, merge/fan-out reasoning, groupby/grain reasoning, null handling, date parsing, validation / reconciliation preparation, Python pipeline review, or performance/memory diagnosis.
Do not use this skill for
Metric definition, DAX measures, SQL transformation logic, readiness approval, dashboard design, or Power BI execution. Those belong to other layers (see boundaries below). Reason up to your edge, then hand off.
Mandatory workflow
- Open
INDEX.md. Do not read the whole knowledge base. - Pick the smallest route that matches the task or symptom.
- Open only the file(s) that route names.
- End on the route's named artifact — a checklist, a JSON pattern set, or an analyzer-style verdict. Never end on an open-ended explanation.
- Stop at boundaries. If the task is metric definition, semantic logic, SQL transformation, or gating, hand off.
SKILL.md -> INDEX.md -> only the named file(s) -> artifact / checklist / verdictWhat this layer owns
Dataframe / pandas / source-prep reasoning and Python pipeline review.
What this layer does NOT own
- Readiness owns stage and gating.
- SQL knowledge owns SQL reasoning, SQL reconciliation, SQL transformation logic.
- Retail KPI knowledge owns metric meaning and contracts.
- DAX knowledge owns measures and semantic-model prerequisites for ready contracts.
- Dashboard design owns visual/page design after metric contracts.
- Execution adapters run things; they never define meaning, mapping, metrics, semantic logic, or approval.
Global stop rules
- Do not clean data before source meaning and grain are understood.
- Do not merge dataframes before join keys and cardinality are known.
- Do not aggregate before declaring grain and additivity.
- Do not infer business meaning from column names alone.
- Do not claim reconciliation passed without evidence.
- Do not fill nulls, coerce dtypes, or join frames without an evidence ledger.
- Do not write production pipeline code in this layer.
- Do not use Python to bypass SQL/readiness validation gates.
- Do not read the whole knowledge base when a router can select files.
- Do not copy book examples or prose. See
references/copyright-safety.md.
Conventions
- ID families and their meaning:
references/id-conventions.md. - Copyright rules for all content:
references/copyright-safety.md. - All examples use the fictional retail schema in
references/retail-dataframe-schema.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.