# Technical Design

> Use when clarified, finalized requirements need converting into a technical design: design.md with integration boundaries, required diagrams, and NFR review, including auth-boundary, tenant-isolation, and application-security-standard mapping checks. This is the technical_design action of the design stretch — one Agent-tool dispatch with subagent_type solution-architect (db-administrator supporting for schema and data-ownership) — distinct from clarify-requirements (which must complete first) and security-architecture-review (the independent security pass that follows).

- Skill: `bhangeef16/technical-design` (Agent Skill)
- Install (CLI): `npx skillmds@latest add bhangeef16/technical-design`
- Raw SKILL.md: https://api.skillmd.com/api/skills/bhangeef16/technical-design/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: BhangeeF16 (https://skillmd.com/u/bhangeef16)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/bhangeef16/technical-design

---


# Technical Design

Convert accepted requirements into the technical design. Nothing here invents business scope, and nothing here starts while a clarification question or a required decision record is open.

## Dispatch

One synchronous `Agent`-tool call, `subagent_type: solution-architect`, with `db-administrator` named in the prompt as the supporting lens for schema design and data-ownership boundaries.

## Required inputs

- `planning.md`.
- finalize-promoted-knowledge output with no unresolved questions.
- `design-system.md` and approved prototypes from detailed-prototypes, when the target has a UI surface in scope.

## Required outputs (all mandatory)

- `design.md`.
- Integration boundaries.
- Required diagrams — at minimum a sequence diagram for any cross-component, asynchronous, or authorization-sensitive flow.
- NFR review with explicit tradeoffs.

## Security checks (required within this action)

- Auth-boundary review.
- Tenant-isolation review.
- Application-security-standard mapping (e.g. OWASP ASVS as an evaluation input).

## Rules — the decision-record gate

- If this change implies a system-wide decision — an architecture change, a migration, a cross-cutting concern — technical design must not begin until an Architecture Decision Record is drafted, reviewed, and its lifecycle status genuinely flipped to Approved, verified by reading the ADR file itself, never accepted from a self-report. Carry the ADR's path and Approved status in this action's own output so every later action can point at the decision rather than re-arguing it.
- A design decision not backed by an existing decision record is requested as one — never buried inline in the design doc.

## Result

Report each required output produced (with its location), the result (pass / pass_with_risks / blocked / failure), and blocking questions as an explicit list (empty list stated explicitly).