paqad-ai 1.4.0 → 1.6.0

package/CHANGELOG.md

@@ -1,5 +1,57 @@
 # paqad-ai
 
+## 1.6.0
+
+### Minor Changes
+
+- [#84](https://github.com/Eliyce/paqad-ai/pull/84) [`82d6c45`](https://github.com/Eliyce/paqad-ai/commit/82d6c457b0768e647cddb932547101e4fe45c4d2) Thanks [@HLasani](https://github.com/HLasani)! - feat([#76](https://github.com/Eliyce/paqad-ai/issues/76)): design-test workflow — UI audit against the project's design-system contract
+
+  Adds a heavyweight design-test workflow that mirrors pentest but grades the running UI against `docs/instructions/design-system/*`. Intent-routed (no slash command), resumable, 9 skills + readiness gate, 7 framework-owned runners under `runtime/scripts/design/`, Playwright live phase, `DT-XXXX` finding ids with stable `token | component | state | a11y | responsive | motion | copy | performance | documentation-drift` categories. Companion `design-retest` workflow preserves IDs across re-runs.
+
+  **Workflow**
+  - New workflow rules: `runtime/capabilities/coding/rules/design-test.md` + `design-retest.md`.
+  - 11 design-test routing triggers at priority 235; 4 design-retest triggers at 245.
+  - `feature-development.yaml` splices design-system reads + per-stage instructions through planning → specification → development → review → checks → documentation_sync (schema-conformant via `merge_mode: append`).
+
+  **Skills (the LLM-reasoning layer)**
+  - `design-system-coverage` (readiness gate, mirrors `stride-threat-model`), `token-conformance-review`, `component-conformance-review`, `state-coverage-review`, `accessibility-review` (WCAG 2.2 A/AA), `responsive-review`, `motion-review`, `copy-and-ia-review`, `design-system-sync`.
+  - **Hard-coded design values** (hex literals, raw `px`/`rem`, ad-hoc font stacks where a token exists) default to **high severity** — first-class findings, not a stylistic preference.
+
+  **Deterministic scripts (the complement to the LLM layer)**
+
+  Per the [agentskills.io](https://agentskills.io) contract: 25 small, focused scripts do the mechanical work so the agent doesn't re-derive it on every run. Each script has `--help`, structured stdout, stderr diagnostics, and meaningful exit codes (0 ok, 1 finding, 2 usage).
+  - design-system-coverage: `count-clauses`, `derive-tier`, `gap-report`
+  - token-conformance-review: `parse-tokens`, `normalize-color`, `match-leak-to-token`
+  - component-conformance-review: `derive-inventory`, `parse-components-md`, `diff-inventories`
+  - state-coverage-review: `extract-source-states`, `extract-tested-states`, `cross-reference-states`
+  - accessibility-review: `static-a11y-scan`, `parse-axe-violations`, `map-axe-to-wcag`
+  - responsive-review: `extract-breakpoints`, `find-horizontal-scroll`, `find-touch-target-violations`
+  - motion-review: `scan-animations`, `parse-motion-budget`, `find-reduced-motion-violations`
+  - copy-and-ia-review: `extract-user-strings`, `check-action-verbs`, `check-terminology`
+  - design-system-sync: `detect-token-additions`, `detect-component-additions`, `propose-tokens-diff`, `propose-components-diff`
+
+  **Framework runners (zero project footprint)**
+
+  `runtime/scripts/design/{scan-tokens, scan-overrides, enumerate-surface, axe-static, coverage, runtime-checks.ts, retest}` — all ship inside paqad-ai; outputs land in the project at `docs/design-test/*` and `.paqad/design-test/runs/*` as work products. Diverges from pentest's project-seeded runner model intentionally; the divergence is documented in the workflow rule.
+
+  **Finding normalizer**
+
+  `DT-` code prefix, design-test category vocabulary, `blocker | nit` severities + `accepted | waived | still-open | needs-manual-verification` statuses.
+
+  **Tests**
+
+  ~200 fixture-driven test cases under `tests/unit/skills/` + `tests/fixtures/design-skills/<skill>/`. The `coverage-completeness` meta-test enforces that every script is referenced by basename in its spec, passes `bash -n`, and that every backticked path in each `SKILL.md` Resources section exists on disk.
+
+  **Skill-authoring rule**
+
+  `docs/instructions/rules/_shared/skill-authoring.md` captures the contract for future skills — anatomy, frontmatter, the deterministic-vs-judgment boundary, the script interface contract, portability workarounds (no `mapfile`, no `\b` in awk, BSD grep alternation quirk, missing-trailing-newline guard), and the testing rules. Auto-loaded by the framework entry, so future skills inherit the contract without being told.
+
+## 1.5.0
+
+### Minor Changes
+
+- [#82](https://github.com/Eliyce/paqad-ai/pull/82) [`f330595`](https://github.com/Eliyce/paqad-ai/commit/f330595d38c5e66afc54443daefe9e4f97a7ea68) Thanks [@HLasani](https://github.com/HLasani)! - Living module lifecycle: prospective decisions, retrospective reconciliation, test-driven health, dashboard drift checks (closes [#80](https://github.com/Eliyce/paqad-ai/issues/80))
+
 ## 1.4.0
 
 ### Minor Changes