<!-- ABOUTME: Workflow for Islamic Value Sensitive Design of provider-mediated software. --> <!-- ABOUTME: Preserves scope, evidence, scholarly, and durable-report boundaries. -->

Islamic Value Sensitive Design

Resources

• Integration contract — load to select standalone, planning, or plan-review mode and enforce handoff freshness.
• Report contract — load before creating or materially updating any report.
• Readable reports — load when composing findings, recommendations, explanations, or decision requests; retain full technical and ethical detail.
• Action routing — load for standalone action selection, menu behavior, and domain lenses.
• Grill-Me alignment — load after standalone action/context routing to resolve material user decisions before substantive analysis.
• Resource index — load only the workflow, evidence, or domain resource selected by the current mode/action.

Rules

• Apply scope boundaries first; I-VSD is provider-responsibility design reasoning, not a fatwa, Sharia/product certification, or proof of ethical outcomes.
• Escalate halal/haram/makrooh/wajib, riba, contested religious-content, and other religious-legal conclusions to qualified Sunni scholarly authority.
• Selected Sunni ethical principles are normative; laws and industry standards may operationalize them only where compatible.
• Review responsibility across strategy, business model, UX, architecture, data/AI, operations, governance, support, portability, and evaluation—not only the surface named in the prompt.
• Trace findings to principles, affected stakeholders, provider-controlled decisions, evidence reviewed, missing evidence, rejected alternatives, and concrete mitigations.
• Persist substantive I-VSD findings under islamic-value-sensitive-design/ partitioned by purpose (governance/ for constitutional policies, consultations/ for domain deep dives, and workstreams/ for dev/active/<task> planning deliverables) using the i-vsd-*.md naming pattern; menus, refusals, context inventories, clarification questions, and agreement prompts remain conversational.
• State uncertainty and evidence limits; never replace absent stakeholder, operational, audit, or scholarly evidence with confidence.

Invocation Modes

• Standalone: with no context, return the action menu. After action/context routing, follow Grill-Me: resolve repository facts directly, recommend one answer, ask one material decision question per response, and continue until the report is aligned or unresolved risks are explicitly deferred. Guided discovery uses its own interview areas with the same one-question decision discipline.
• Planning: consume the planner's shared evidence packet, create a draft workstream report, allow material questions, then revalidate the completed plan/task mapping before declaring it plan-aligned.
• Plan review: bind the moral review to exact plan/I-VSD revisions; a material CTO rewrite makes the report stale and requires revalidation.

Follow integration-contract.md. Load only the routed resources, never the full library.

Output contract

Follow report-contract.md for identity, metadata, stable finding/mitigation IDs, lifecycle, evidence levels, and required headings. Planning and plan-review reports also include its Planning Handoff.

Reports and chat explain practical meaning before the technical/ethical reasoning, following readable reports. Add explanations rather than shorten away terminology, mechanisms, evidence, or uncertainty.

The chat response names the report and gives a self-contained recommendation, rationale, material limits, and any decision needed. The report retains the complete analysis and traceability; its path is not a substitute for an understandable answer.

Verification

• Confirm the request is provider-mediated software responsibility and not a religious ruling.
• Confirm no-context standalone invocation still returns actions without creating a report.
• Confirm the workflow can stop for material context in every mode.
• For substantive output, confirm a Markdown report was created or updated at the mapped path; for a routing response, confirm no report was written.
• Check report identity, status, disposition, evidence revision, stable finding IDs, and every authority boundary.
• Check that summaries and finding explanations are understandable without decoding tables, while the complete technical, ethical, and evidence fields remain intact.
• For planning/review, confirm every material IVSD-* ID maps to a scenario/task, explicit non-applicability, or escalation gate and the reviewed revision is current.
• Run host-repository link, schema, and whitespace checks when this skill or its resources change.