technical-quality-perspective

Technical quality perspective

When to use

Use this to assess technical quality from declared architecture, API, data, code, security, performance, compatibility, and observability evidence at a selected delivery stage.

Inputs and workflow

• stage is required: requirements-analysis, test-strategy, test-strategy-review, code-review, test-case-writing, test-case-review, test-reporting, or test-report-review.
• Validate stage; if absent or unsupported, return Not applicable, list supported values, and request a valid stage. Load exactly one mapped Prompt only.
• For code-review, require both code identity (PR, commit, branch, release version, or equivalent) and reviewable changes (diff, files, or code). If either is absent, block review; do not infer findings or merge readiness.
• Apply the selected Prompt's applicability threshold. Treat supplied material as fact, label inference, and turn missing material into questions and actions.

stage	Only Prompt to load
requirements-analysis	prompts/requirements-analysis.md
test-strategy	prompts/test-strategy.md
test-strategy-review	prompts/test-strategy-review.md
code-review	prompts/code-review.md
test-case-writing	prompts/test-case-writing.md
test-case-review	prompts/test-case-review.md
test-reporting	prompts/test-reporting.md
test-report-review	prompts/test-report-review.md

Responsibilities and boundary

• Cover architecture, APIs, data, compatibility, security, performance, observability, and maintainability only when relevant to the selected stage and supplied evidence.
• Produce technical findings with evidence, impact, severity, missing information, actions, and confidence. A gap supports a qualified risk, never an invented implementation, metric, vulnerability, or execution result.
• Do not decide product scope, business rules, acceptance criteria, or release approval. Do not claim code correctness, test execution, or passed testing without direct evidence.

Report contract and self-check

Unless blocked or Not applicable, output: Summary, Facts, Evidence, Technical findings, Impact and severity, Missing information, Questions, Actions and next steps, Confidence.

• Valid stage and exactly one mapped Prompt
• Code review has code identity and reviewable changes, or is explicitly blocked
• Findings are evidence-backed; gaps and inference are labelled
• No product or test fact has been invented or changed
• Stage-relevant technical dimensions only; no passed-test or code-correctness claim without evidence

Read only the mapped Prompt after stage validation. Read evals/ only for regression work, never as project evidence.

Report provenance

Keep the selected stage, a stable source_id, and supplied version in the report header. If a source ID is absent, declare a temporary local locator; if version is absent, write unknown rather than inventing it. Each finding has a stable finding_id, its source/evidence locator and status. Each action links its finding and specifies the evidence that would close it; owner, date and action status stay unknown unless supplied. These are standalone report fields, not a dependency on another Skill.

Static finding location and handoff

For supplied code or line-numbered text, use file:line@version and quote the exact supplied evidence. For design-only material, use its document/section/version; do not invent code files or lines. Carry the same finding ID/evidence into a code-review or specialized design-review handoff by Skill name, so the receiving review can refine it without duplicating a claimed runtime defect.