report-intent-interview
Skill2 starsChanged 4 months agoApache-2.0View on GitHub
Given approved metric contracts and a ready semantic model, interview the report requester to capture what a dashboard report is FOR -- audience, purpose, supported decision, review cadence, primary business questions, outcome/driver/guardrail metrics (referenced by name only, never defined), comparisons, dimensions/filters, expected actions/exceptions, pages/drill paths, and mobile/accessibility/language/RTL needs -- and record the result as a committed Report Intent artifact plus a named-human `report_intent_approval` decision. Use when someone describes a report they need in plain language, before any page blueprint, visual spec, or report composition is authored. It asks obvious low-risk items as one batch and critical items individually with explicit named-human approval; it masks PII by default; it records the intent and STOPS. It does NOT define metric meaning, does NOT self-grant any approval, and does NOT emit a confidence score.
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
- 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-python-knowledgeSkill
- bi-sql-knowledgeSkill
- retail-kpi-knowledgeSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 3 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 40
Before dashboard-design, the dashboard coordinator, page blueprints, visual specs, or report composition -- those stages trace their …
- Very specific: line 76
Reference metrics by name only: Every outcome_metrics / driver_metrics / guardrail_metrics entry names an approved metric contract (name + …
- Very specific: line 86
Record the artifact and the decision: Write mappings/<subject-area>/design/report-intent.yaml per templates/report-intent.yaml, and record …
- Very specific: line 104
Never self-grant report_intent_approval (Principle V.
- Distinctive: line 106
Never advance a readiness stage.
- Very specific: line 107
Never emit a numeric confidence score (never_fabricate_a_confidence_score).
- Very specific: line 108
Never define metric meaning here (route to retail-kpi-knowledge / the metric-contract layer).
- Distinctive: line 110
Never display raw suspected-PII values by default, and never commit them.
- Distinctive: line 111
Never invent a metric contract to make a reference resolve.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
report-intent-interview
The Report Intent Interview is the missing front end of the dashboard
intelligence journey (spec 123, US1). It converts a conversational report
request into a committed, reviewable report-intent.yaml so every downstream
artifact (blueprint, composition, preview, audit, PBIR) traces back to a clear,
approved report purpose instead of an unrecorded assumption.
This skill mirrors .claude/skills/business-knowledge-interview/SKILL.md
structurally (load-existing-first, batch/critical question grouping, PII
masking, record + STOP, never self-grant) but does NOT extend it: the
required_inputs and focus differ. Behavior is governed by the contract
contracts/interview/report-intent-interview.yaml; the shape of the produced
artifact is templates/report-intent.yaml. The static rule DL9
(seshat check) verifies the recorded artifact's SHAPE, not the conversation.
When to use
- Someone describes a report they need in plain language ("I need a weekly report for branch managers to spot underperforming branches").
- Before
dashboard-design, the dashboard coordinator, page blueprints, visual specs, or report composition -- those stages trace theirbusiness_questionback to a question declared here (FR-002a) and are blocked until this intent's ownreport_intent_approvaldecision is approved.
Preconditions (stop if unmet)
- Approved metric contracts exist for the subject area (
readiness.status: pass). No approved contract for a metric the request needs => record a gap, route upstream to metric-contract definition, and leave the intent blocked on that metric (FR-004) -- never define the metric here. - The semantic model is ready (
semantic_model_ready: pass). Not ready => stop, name the missing readiness, and do not proceed to design. - Load the existing Report Intent (if any) first. An existing intent is presented for confirmation or supersession -- never silently overwritten.
How it runs
- Ground every question in the request and the approved contracts. Focus on: audience, purpose, supported decision, review cadence, primary business questions, metric roles by name (outcome/driver/guardrail), comparisons, dimensions and filters, expected actions and exceptions, pages/drill paths, mobile/accessibility/language/RTL needs, and exclusions/non-goals.
- Refuse to commit a vague request. "Build an executive dashboard" gets
focused disambiguating questions; the intent is NOT committed until
audience,purpose, and at least one primarybusiness_questionsentry are resolved (US1 AC#4). - Batch the obvious low-risk items into one owner confirmation (e.g. comparison phrasing, filter defaults). Critical decision types are never in a batch. The owner may exclude any item; the remainder is approved in one action and each excluded item becomes an individual pending question.
- Ask critical decisions individually. The intent's own approval
(
report_intent_approval) is a critical decision type, authority classreport_owner(contracts/knowledge/approval-authority.yaml); it requires an explicit, per-decision approval from a named human -- the agent never self-grants it (Principle V). - Reference metrics by name only. Every
outcome_metrics/driver_metrics/guardrail_metricsentry names an approved metric contract (name+store_ref+status_required: pass) -- the same triple shape asdashboard-page-blueprint.yaml'srequired_metric_contracts. There is no formula/DAX field. A metric with no approved contract is a gap: record it, route it upstream, and leave the intentblockedon that metric -- never invent the metric here. - Mask suspected PII by default. Any PII-adjacent dimension/filter follows
the shipped masking behavior; unmasking requires an explicit owner
instruction, recorded as a
pii_handlingdecision. - Record the artifact and the decision. Write
mappings/<subject-area>/design/report-intent.yamlpertemplates/report-intent.yaml, and record thereport_intent_approvaldecision in the project Decision Store once a named human of authority classreport_ownerapproves. Unanswered items staypending/needs_user_input; a metric gap staysblockedwith a matchingblocking_reasonsentry.
Exit
Report the current gate verdict for the report_intent stage
(contracts/report/report-intent.yaml) and the next allowed stage
(dashboard_blueprint) -- which stays blocked until this intent's
report_intent_approval is valid (an unapproved intent yields blocked, never
a false pass).
Hard stops (never)
- Never self-grant
report_intent_approval(Principle V;never_self_grant_approval). - Never advance a readiness stage.
- Never emit a numeric confidence score (
never_fabricate_a_confidence_score). - Never define metric meaning here (route to
retail-kpi-knowledge/ the metric-contract layer). - Never display raw suspected-PII values by default, and never commit them.
- Never invent a metric contract to make a reference resolve; record the gap and leave the intent blocked instead (FR-003/FR-004).
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.

