From 020a357e598d1c72e4dbcf717d24d9e00d4466a2 Mon Sep 17 00:00:00 2001 From: Arda <233561772+ardamoustafa1@users.noreply.github.com> Date: Fri, 17 Jul 2026 11:00:37 +0300 Subject: [PATCH] Polish launch proof and contribution templates --- .github/ISSUE_TEMPLATE/independent-eval.md | 24 ++++++ .github/ISSUE_TEMPLATE/showcase-submission.md | 26 +++++++ .github/PULL_REQUEST_TEMPLATE.md | 3 + LAUNCH.md | 18 +++-- README.md | 11 +-- evals/README.md | 7 ++ evals/RUN_TEMPLATE.md | 77 +++++++++++++++++++ package.json | 2 +- 8 files changed, 154 insertions(+), 14 deletions(-) create mode 100644 .github/ISSUE_TEMPLATE/independent-eval.md create mode 100644 .github/ISSUE_TEMPLATE/showcase-submission.md create mode 100644 evals/RUN_TEMPLATE.md diff --git a/.github/ISSUE_TEMPLATE/independent-eval.md b/.github/ISSUE_TEMPLATE/independent-eval.md new file mode 100644 index 00000000..204d90a8 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/independent-eval.md @@ -0,0 +1,24 @@ +--- +name: Independent eval run +about: Submit or coordinate a third-party DesignOS benchmark run +title: "[eval] Run 00X — " +labels: eval, evidence +--- + +## Run summary + + +## Protocol +- [ ] Same model/settings in control and treatment +- [ ] Same brief text in both arms +- [ ] No follow-up steering unless declared +- [ ] Outputs randomized before judging +- [ ] `evals/judge-prompt.md` used for blind scoring +- [ ] Validator raw output attached +- [ ] Raw generated artifacts attached or linked + +## Report + + +## What should change because of this? + diff --git a/.github/ISSUE_TEMPLATE/showcase-submission.md b/.github/ISSUE_TEMPLATE/showcase-submission.md new file mode 100644 index 00000000..74928387 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/showcase-submission.md @@ -0,0 +1,26 @@ +--- +name: Showcase submission +about: Submit a real project built with DesignOS +title: "[showcase] " +labels: showcase +--- + +## Project +- Live URL: +- Repository or case-study URL: +- Sector/type: + +## What DesignOS changed + + +## Proof +- [ ] The page/app is publicly reachable +- [ ] The claim does not invent users, metrics, testimonials, or badges +- [ ] You are allowed to share this work +- [ ] You are comfortable with the project being linked from `SHOWCASE.md` + +## Suggested row + +```markdown +| [Project name](https://live-url) | sector/type | one honest sentence on what DesignOS changed | +``` diff --git a/.github/PULL_REQUEST_TEMPLATE.md b/.github/PULL_REQUEST_TEMPLATE.md index 8fa72193..0cb9fbda 100644 --- a/.github/PULL_REQUEST_TEMPLATE.md +++ b/.github/PULL_REQUEST_TEMPLATE.md @@ -7,3 +7,6 @@ - [ ] Cross-references resolve (reference check script run) - [ ] Kernel routing table updated (if a module was added) - [ ] README structure tree updated (if the tree changed) +- [ ] `npm test` run, or reason stated +- [ ] `npm run validate` run, or reason stated +- [ ] Any proof/eval claim names who ran it and links raw outputs diff --git a/LAUNCH.md b/LAUNCH.md index 1652fdda..2e7be9f4 100644 --- a/LAUNCH.md +++ b/LAUNCH.md @@ -4,22 +4,24 @@ Stars come from distribution × proof × timing — not from file count. This is operational plan for taking DesignOS worldwide. Work it top to bottom; check items off in your fork. -## Phase 0 — Pre-flight (do BEFORE any public link exists) -- [ ] Replace every `USER` placeholder (README, website, package.json repo fields) -- [ ] Push to GitHub; enable Pages on `website/` (Settings → Pages → /website folder or - a `gh-pages` action); verify the live URL on mobile +## Phase 0 — Pre-flight (do BEFORE the big public push) +- [x] Replace every `USER` placeholder (README, website, package.json repo fields) +- [x] Push to GitHub; enable Pages via GitHub Actions; verify the live URL on mobile. + Last checked: returned HTTP + 200 on 2026-07-17. - [ ] Repo polish: description line ("Design intelligence OS for AI coding agents — one prompt in, Stripe-grade UI out"), topics (`claude`, `claude-code`, `cursor`, `ai-agents`, `design-systems`, `prompt-engineering`, `ui-design`), social-preview - image uploaded — `press/social-preview.png` (1024×1024, dark, the boot-log + image uploaded — `press/social-preview.png` (1280×640, dark, the boot-log terminal as the visual) is ready at Settings → Social preview → Edit - [ ] Run the full validator suite + click every README link - [ ] **Record the 30-second GIF** (the single highest-leverage asset): screen capture of a real session — prompt in → boot log → modules routing → scorecard → the showcase page. Put it at the top of the README. Tools: any recorder + gifski. No GIF = half the launch. -- [ ] Run `evals/` on at least one model pair and put the numbers in the README — - "measured, not vibed" is the differentiator claim; back it before HN asks +- [ ] Get Run 003 from an external user with [`evals/RUN_TEMPLATE.md`](evals/RUN_TEMPLATE.md) + attached — "measured, not vibed" is the differentiator claim; maintainer-only + numbers are useful instrumentation, not proof ## Phase 1 — Seeding (1–2 weeks before the spike) - [ ] Soft-share in 3–5 niche communities where feedback > reach: Claude Code Discord, @@ -62,7 +64,7 @@ in your fork. each has a submission path; the GIF + one-paragraph pitch - [ ] Respond to every issue within 24h for the first month — early responsiveness converts visitors to contributors; slow repos die in the trough -- [ ] Ship a visible v1.3/v1.4 within 3 weeks of launch (the changelog IS marketing — +- [ ] Ship a visible v2.1 within 3 weeks of launch (the changelog IS marketing — `patterns/changelog.md`, applied to ourselves) ## The message discipline (all channels, one voice) diff --git a/README.md b/README.md index 0daaa4b7..41cc6183 100644 --- a/README.md +++ b/README.md @@ -13,7 +13,7 @@ [![GitHub stars](https://img.shields.io/github/stars/ardamoustafa1/DesignOS?style=flat&color=10b981)](https://github.com/ardamoustafa1/DesignOS/stargazers) [![License: MIT](https://img.shields.io/badge/License-MIT-black.svg)](LICENSE) -[![Version](https://img.shields.io/badge/version-2.0.1-blue.svg)](CHANGELOG.md) +[![Version](https://img.shields.io/badge/version-2.0.2-blue.svg)](CHANGELOG.md) [![Live proof](https://github.com/ardamoustafa1/DesignOS/actions/workflows/proof.yml/badge.svg)](https://github.com/ardamoustafa1/DesignOS/actions/workflows/proof.yml) [![Validate](https://github.com/ardamoustafa1/DesignOS/actions/workflows/validate.yml/badge.svg)](https://github.com/ardamoustafa1/DesignOS/actions/workflows/validate.yml) [![WCAG 2.2 AA](https://img.shields.io/badge/A11y-WCAG%202.2%20AA-2ea44f.svg)](checklists/accessibility.md) @@ -66,7 +66,7 @@ The agent stops thinking *"I'll make a button"* and starts thinking **We're trying to measure this, not just claim it.** Below is a maintainer-run sanity check — **not independent validation** ([full caveats](evals/RESULTS.md)) — showing the validators catch what they're supposed to catch. The eval slot that would actually prove -the claim is open and unfilled; [run it and PR your numbers](evals/RESULTS.md#run-002--your-model-your-judge-pending--this-is-the-one-that-matters). +the claim is open and unfilled; [run it and PR your numbers](evals/RESULTS.md#run-003--your-model-your-judge-the-one-that-matters-most). | Mechanical check (same brief) | Default-style output | Through DesignOS | |---|---:|---:| @@ -231,8 +231,9 @@ Every page has its decision paper-trail: miss, an unverified metric, a CSS cascade bug found at 577px). 🔬 **[Measure it yourself](evals/README.md)** — 10 fixed briefs, a paste-ready -[neutral judge prompt](evals/judge-prompt.md), validators as the objective floor, and -[published results](evals/RESULTS.md) with the caveats stated plainly. +[neutral judge prompt](evals/judge-prompt.md), validators as the objective floor, a +[run report template](evals/RUN_TEMPLATE.md), and [published results](evals/RESULTS.md) +with the caveats stated plainly. 🌐 **[The website](website/index.html)** — designed by the system it documents. View source: every value resolves to a token. @@ -278,7 +279,7 @@ Under threshold → back to the loop with specific objections, *before you ever - 🏗 **[Add your work to the Showcase](SHOWCASE.md)** — and take the [![badge](https://img.shields.io/badge/designed%20with-DesignOS-10b981)](SHOWCASE.md) with you - 🥊 **[Challenge a rule](CONTRIBUTING.md)** — every rule stands on evidence; bring yours -- 🔍 **[Run the missing independent eval](evals/RESULTS.md)** — the single highest-value +- 🔍 **[Run the missing independent eval](evals/RUN_TEMPLATE.md)** — the single highest-value contribution available right now - 💬 **[Discussions](DISCUSSIONS.md)** — questions, show-and-tell, module ideas (setup guide; enable in repo Settings) - 🗺 **[Roadmap](ROADMAP.md)** · 🚀 **[Launch playbook](LAUNCH.md)** · 🛡 **[Security](SECURITY.md)** · 📰 **[Press kit](press/README.md)** · ⚠️ **[Known limitations](LIMITATIONS.md)** · 🏢 **[Enterprise](ENTERPRISE.md)** · ⚖️ **[Governance](GOVERNANCE.md)** diff --git a/evals/README.md b/evals/README.md index 531f45e5..b74ff35c 100644 --- a/evals/README.md +++ b/evals/README.md @@ -55,3 +55,10 @@ After editing modules: re-run the affected briefs (a pricing-module change → b and 8). Score deltas below the noise floor are fine; systematic drops are a rule-challenge issue (`CONTRIBUTING.md`) with data attached — exactly the evidence standard we ask of contributors. + +## Submitting Run 003+ + +Use [`RUN_TEMPLATE.md`](RUN_TEMPLATE.md) for independent reports. The template is +intentionally specific: model versions, agent surface, prompt text, raw validator output, +judge-pass medians, and negative findings all belong in the PR. A mixed or negative run +is useful evidence; a perfect-looking report with no raw outputs is marketing, not eval. diff --git a/evals/RUN_TEMPLATE.md b/evals/RUN_TEMPLATE.md new file mode 100644 index 00000000..ad438861 --- /dev/null +++ b/evals/RUN_TEMPLATE.md @@ -0,0 +1,77 @@ +# Independent Eval Run Template + +Copy this file into a PR section or attach it as `evals/runs/run-00X.md`. +Reports can be positive, mixed, or negative. The only bad report is one that hides +conditions or raw outputs. + +## Run metadata + +| Field | Value | +|---|---| +| Run ID | Run 003 | +| Date | YYYY-MM-DD | +| Runner | name / GitHub handle | +| Agent surface | Claude Code / Cursor / Copilot / Windsurf / Cline / Aider / other | +| Model(s) | exact model names and versions, if visible | +| OS / browser | environment details | +| Brief set | all 10 / declared subset | +| Time budget | per arm | +| Follow-up steering | none / explain deviation | + +## Installation notes + +- Control arm setup: +- Treatment arm setup: +- DesignOS commit SHA: +- Any local modifications: + +## Validator results + +Run the same commands for both arms and paste the raw output or a link to the raw log. + +```bash +node validators/check-drift.js path/to/control-output.html +node validators/check-a11y-basics.js path/to/control-output.html +node validators/check-drift.js path/to/treatment-output.html +node validators/check-a11y-basics.js path/to/treatment-output.html +``` + +| Brief | Arm | Drift findings | A11y findings | Contrast notes | +|---|---|---:|---:|---| +| B-001 | Control | | | | +| B-001 | Treatment | | | | + +## Blind judge results + +Use `evals/judge-prompt.md`. Run three judge passes when using a model judge and report +medians, not just the best pass. + +| Brief | Arm | Hierarchy | Craft | Accessibility | States | Distinctiveness | Fitness | Median total | +|---|---|---:|---:|---:|---:|---:|---:|---:| +| B-001 | A/B label | | | | | | | | + +## Raw artifacts + +- Control outputs: +- Treatment outputs: +- Judge transcripts: +- Validator logs: + +## Findings + +### Where DesignOS helped + +- + +### Where DesignOS hurt or added cost + +- + +### Ambiguous results + +- + +## Conclusion + +State the result in one honest paragraph. Avoid universal claims. This run measures +this model, this surface, this brief set, under these conditions. diff --git a/package.json b/package.json index 18589539..ba2fdc3b 100644 --- a/package.json +++ b/package.json @@ -1,7 +1,7 @@ { "name": "designos", "version": "2.0.2", - "description": "The Design Intelligence Operating System for AI coding agents — Claude, Fable, Cursor, Copilot.", + "description": "The Design Intelligence Operating System for AI coding agents — Claude, Cursor, Copilot, Windsurf, Cline, Aider.", "author": "Arda Moustafa (https://github.com/ardamoustafa1)", "repository": { "type": "git",