Skip to content

Releases

v1.1.0: the website and the redesigned app

Choose a tag to compare

@ved-cmpt ved-cmpt released this 12 Nov 04:27

A website and a redesigned app. Behaviour, the API and the data are unchanged.

Run 07 (2026-11-11 to 2026-11-12)

  • Brand. An original mark (a block of legacy code with a new shoot growing from it), an outlined wordmark, lockups, app and maskable icons, a favicon set and an Open Graph image, as SVG with PNG and ICO renderings and the scripts that rebuild them (docs/BRAND.md, ADR-051).
  • Design system. Dark-first SCSS tokens with a light theme; self-hosted, subset Inter and JetBrains Mono (SIL OFL 1.1); spacing, radii, shadows and motion tokens; every animation stops under prefers-reduced-motion.
  • Website at /. A sticky navbar; a hero that replays the stored RATECALC run (484 of 500 cases, blocked, then 500 of 500 after one repair); the problem and an honest comparison; the seven steps lit up on scroll; four interactive demos built from real BillRite code and stored results (the rounding defect, the two-digit year window, the meter rollover and the blocked prompt injection), each with a link into the same run in the app; the reference deployment model, labelled as a model with what v1 builds; features, security, the evaluation (mock and live columns), open source, an FAQ and a footer. Its numbers come from frontend/public/data/landing-snapshot.json, written from stored runs by scripts/landing_snapshot.py with run ids and the seed, checked in CI and against a fresh seed (ADR-054). SEO and social tags, robots.txt, sitemap.xml and a web manifest.
  • The app moved to /app. Old paths redirect; the session is restored by the route guards, so the website never waits for it; links into the demo by name, such as /app/open/module/RATECALC (ADR-052).
  • New app shell. A collapsible icon sidebar with a project switcher and a user menu, breadcrumbs, a command palette (Ctrl or Cmd+K) over screens, modules, rules, sources and actions, G-letter shortcuts with a help sheet (?), toasts, skeleton loaders, empty states and a dismissible first-visit tour of the rounding story.
  • Every screen redesigned. Sign-in with one-click demo accounts; project cards; an overview with KPI cards, a module-state chart, a coverage and mutation chart and an activity timeline; an inventory with a call graph that pans, zooms, fits and switches layout; syntax-coloured COBOL and C#; rules with state pills, a table view and links to one rule; a plan with a service map and a timeline; module list bars; a module workspace with a result hero, COBOL beside C# with what the repair changed, field-level diffs, a coverage map, a mutation gauge, a gate checklist and a facade mode track; a report with a contents rail; an audit log with category chips and a timeline view. Still no highlighting, graph or chart library (ADR-053).
  • Accessibility. axe finds no serious or critical issue on any page, in either theme, at 375, 768, 1280 or 1440 px.
  • Performance. The origin gzips static text (never API responses), and the landing page no longer shifts as it loads (ADR-055). Lighthouse on the live website: performance 91 on mobile and 100 on desktop, with accessibility, best practices and SEO at 100 on both.
  • Fixed: the upload button's mis-encoded ellipsis; the README's rounding example now names the stored case (December, case 82).
  • Tests: pytest 788, Angular 209 (from 163), .NET 26, browser journeys and accessibility 32 (from 18), smoke 16 (from 14), all passing; the smoke checklist passes against the live service.

Website and live demo: https://offshoot-go2y.onrender.com (the app is under /app). Full evidence: docs/RUN_REPORTS/RUN_07.md.

Offshoot v1.0.0

Choose a tag to compare

@ved-cmpt ved-cmpt released this 08 Nov 20:39

Offshoot: AI-assisted COBOL → C# modernization, with proof.

Live demo: https://offshoot-go2y.onrender.com. Sign in as viewer, engineer, architect or admin, all with the password demo1234. It is a free instance, so the first visit after 15 idle minutes takes about 45 seconds.

What is in v1.0.0

  • Deterministic inventory of a synthetic 1990s COBOL billing system (BillRite): copybooks, call graph, data layouts, dead code.
  • AI rule extraction with every cited span and data name checked by the server. A person verifies each rule; the model can never approve anything.
  • A modernization plan, then migration to C# in a network-less sandbox that forbids processes, network, file system and reflection, with a bounded repair loop.
  • Golden-master equivalence. The real COBOL and the new C# run on the same 500 seeded inputs, compared field by field. Legacy paragraph coverage and Stryker.NET mutation testing back it up, and server-side gates decide.
  • Human approval and a strangler-fig facade (YARP) with legacy, shadow and modern routing per module.
  • A dashboard, an assessment report and an append-only audit log.
  • A public demo that never runs code: it shows stored runs of the real sandbox, and resets itself daily.

The headline story

The first C# translation of the rate calculator rounded half a cent to the even cent; COBOL rounds it up. Its own unit tests passed, but the harness found 16 of 500 cases off by exactly one cent and blocked the module. One repair later, 500 of 500 matched, and the module went through shadow mode (20 requests, 0 mismatches) to modern.

Measured

  • Mock pipeline: rule precision 95.65% and recall 88.00%; 4 of 4 modules compiled; 3 of 4 were equivalent on the first attempt and 4 of 4 after one repair.
  • Live Gemini sample (gemini-3.5-flash, 20 calls under a hard cap of 30):
    • Rule precision 93.33% and recall 63.64% on two programs.
    • The live RATECALC migration compiled and matched 500 of 500 cases on the first attempt.
    • The first 17 calls failed, and that is recorded. Gemini refused the structured response schema, and the adapter's retries then hit the rate limit.

Details are in docs/EVALUATION.md.

Quality

pytest 787, Angular 163, .NET 26, browser and accessibility suite 18, smoke checklist 14 (also against the live demo). CI checks every push; the slow suites run weekly.

See CHANGELOG.md for the full history and docs/RUN_REPORTS/ for what each of the six runs did and proved.