Make repo pages distinct and sitemap dates honest; plain-language copy

AGENTS.md

@@ -66,7 +66,7 @@ app/
   twitter-image.tsx                         # next/og convention — twitter:image, re-exports opengraph-image (auto-wired)
   og/repo/[...slug]/route.tsx               # per-repo OG image — a plain route, because the next/og file
                                             # convention would sit at a path the catch-all page route also claims
-  score/page.tsx                            # Live Score entry — URL form, past scores, FAQ
+  score/page.tsx                            # Live Score entry — URL form, visitor's recent scores + top leaderboard repos, FAQ
   score/opengraph-image.tsx                 # next/og convention — Live Score OG image (auto-wired)
   score/twitter-image.tsx                   # next/og convention — /score twitter:image, re-exports (auto-wired)
   score/[host]/[owner]/[name]/page.tsx      # live score; result cached 1h per repo (unstable_cache); redirects to the repo page when indexed
@@ -81,7 +81,7 @@ components/               # Tailwind-styled React components
   HostPill.tsx, HostSelect.tsx, Medal.tsx, ModelPills.tsx,
   MobileNav.tsx, Pagination.tsx, SearchBar.tsx, SelectMenu.tsx, SortSelect.tsx,
   SignalRow.tsx, SuggestionItem.tsx, VersionPill.tsx,
-  RepoHero.tsx, ScoreDeltaPopover.tsx, SignalListCard.tsx, ModelSuggestions.tsx, PerModelScores.tsx,
+  RepoHero.tsx, RepoSummary.tsx, ScoreDeltaPopover.tsx, SignalListCard.tsx, ModelSuggestions.tsx, PerModelScores.tsx,
   AlternativesStrip.tsx, BreadcrumbJsonLd.tsx, HomeJsonLd.tsx, ExternalLink.tsx,
   BadgeEmbed.tsx, ActionEmbed.tsx, PeerlistCard.tsx, PeerlistBadge.tsx, ProductHuntBadge.tsx,
   CopySnippet.tsx, PackageLookupForm.tsx,
@@ -99,6 +99,7 @@ lib/
     badge.ts              # SVG badge renderer (used by /api/badge)
     contact.ts            # packageRequestIssueUrl — pre-filled GitHub issue link for unscored packages
     repo-path.ts          # repoPath / repoIdentity — the one place a repo URL is built or parsed
+    repo-summary.ts       # per-repo rank, best/worst agent, key signals — the repo-specific text on repo pages
   scoring/
     signals/              # one file per signal + helpers + types + index
     weights.ts            # per-model weight tables
@@ -133,6 +134,7 @@ tests/
   parse-repo-url.test.ts  # GH / GL / BB parsing + edge cases
   scorer.test.ts          # scoreRepo, topImprovements
   badge-adoption.test.ts  # detectBadgeEmbed — README badge-embed detection
+  repo-summary.test.ts    # summarizeRepo / keySignals — ranks, ties, best/worst agent
   path-resolution.test.ts # firstExisting / resolveRelative / resolveAllRelative — case-insensitive lookup
   live-score.test.ts      # content-candidate coverage vs the signals, path traversal, host URLs
   signals/                # one *.test.ts per signal
@@ -210,7 +212,7 @@ If either sibling isn't present locally, flag it; never silently skip the propag
 
 1. Add a `ModelProfile` to `MODELS` in `lib/scoring/weights.ts` — weights for every signal.
 2. Appears automatically in the leaderboard model pills, methodology weight-profile panel, and repo-page suggestions.
-3. Update the hard-coded agent lists/counts that **don't** derive from `MODELS`: `APP_DESCRIPTION` + `APP_KEYWORDS` in `lib/version.ts`, `lib/skill-content.ts`, the "Which agents" FAQ in `app/methodology/page.tsx`, `app/page.tsx`, `app/skill/page.tsx`, `app/action/page.tsx`, `app/terms/page.tsx`, `app/opengraph-image.tsx`, the footer strip in `app/og/repo/[...slug]/route.tsx`, the `generateMetadata` description + JSON-LD description in `app/repo/[...slug]/page.tsx`, the Dataset description in `components/HomeJsonLd.tsx`, and `README.md`. Grep the current count word (e.g. `eight`) and the trailing `OpenHands, Pi` to find them all — `tasks/` and `lib/changelog.ts` are historical records and stay as-shipped, and `.claude/skills/agent-friendly/SKILL.md` is an install artifact pinned by `skills-lock.json` (it refreshes via `npx skills add`, never by hand).
+3. Update the hard-coded agent lists/counts that **don't** derive from `MODELS`: `APP_DESCRIPTION` + `APP_KEYWORDS` in `lib/version.ts`, `lib/skill-content.ts`, the "Which agents" FAQ in `app/methodology/page.tsx`, `app/page.tsx`, `app/score/page.tsx`, `app/skill/page.tsx`, `app/action/page.tsx`, `app/terms/page.tsx`, `app/opengraph-image.tsx`, the footer strip in `app/og/repo/[...slug]/route.tsx`, the Dataset description in `components/HomeJsonLd.tsx`, and `README.md`. Grep the current count as a word and a digit (e.g. `nine`, `9`) and the trailing `OpenHands, Pi` to find them all — `tasks/` and `lib/changelog.ts` are historical records and stay as-shipped, and `.claude/skills/agent-friendly/SKILL.md` is an install artifact pinned by `skills-lock.json` (it refreshes via `npx skills add`, never by hand).
 4. Existing repos keep their old per-model rows until rescored. `PerModelScores` renders the missing model as "—" (not 0), but **everything that reads `model_score` via a JOIN degrades silently to empty** until the backfill lands: the leaderboard (`listLeaderboard`, i.e. `/?model=<new-id>`) and `getAlternatives` (the repo page's alternatives strip under `?model=<new-id>`) both inner-join `model_score` and return **zero rows** for a model with no rows yet — no error, just an empty page. Reweighting an existing model has the same staleness problem in reverse: stored scores stay at the old weights until rescored. Either trigger `scheduled-rescore.yml` via `workflow_dispatch` at deploy time, or accept up to 6 hours of the empty/stale state until the cron runs.
 5. **Mirror to both siblings**: copy the weights change into `../agent-friendly-action/src/scoring/weights.ts` **and** `../agent-friendly-skill/src/scoring/weights.ts`, and log under "Unreleased" in each sibling's `CHANGELOG.md`.
 

README.md

@@ -193,7 +193,7 @@ GitLab and Bitbucket are implemented and score identically to a clone, but ship
 
 ## Companion: PR-diff GitHub Action
 
-[`hsnice16/agent-friendly-action`](https://github.com/hsnice16/agent-friendly-action) runs the same scorer inside your CI and posts a per-PR score-delta comment — _"this PR drops your Claude Code score by 4.1 points because it removed CI config."_ Opt-in via an `AGENTS_BADGE_TOKEN` secret; falls through silently when unset. Each repo detail page on the dashboard ships a copy-paste workflow snippet under "Catch score regressions on every PR".
+[`hsnice16/agent-friendly-action`](https://github.com/hsnice16/agent-friendly-action) runs the same scorer inside your CI and posts a per-PR score-delta comment — _"this PR drops your Claude Code score by 4.1 points because it removed CI config."_ Opt-in via an `AGENTS_BADGE_TOKEN` secret; falls through silently when unset. Each repo detail page on the dashboard ships a copy-paste workflow snippet under "Check the score on every pull request".
 
 ## Companion: agent skill
 
@@ -204,7 +204,7 @@ GitLab and Bitbucket are implemented and score identically to a clone, but ship
 Read-only JSON endpoints for external integrators (skills, hooks, browser overlays, third-party tools):
 
 - `GET /api/score?host=<host>&repo=<owner>/<name>` — look up an indexed repo by host + owner/name. Returns `{ repo, signals, modelScores }` on 200; `{ error: "not_indexed" }` with status 404 when the repo isn't in our DB. The natural lookup endpoint for any tool that has a repo URL but not our internal id.
-- `GET /api/repos` — full leaderboard (id, owner, name, host, stars, overall_score, per-model scores).
+- `GET /api/repos` — full leaderboard (each repo row plus its per-model scores).
 - `GET /api/repo/<id>` — per-repo detail (signals, model scores, top improvements). Requires the internal id; use `/api/score` first if you only have host + owner/name.
 - `GET /api/badge/<host>/<owner>/<name>.svg` — embeddable SVG badge. `?model=<id>` for per-model variants.
 - `GET /api/package/<registry>/<name>` — resolve npm / PyPI / Cargo package → source-repo score (or `unresolved` when the registry doesn't expose a repo URL).

app/about/page.tsx

@@ -17,7 +17,7 @@ export const metadata: Metadata = {
     url: "/about",
     type: "article",
   },
-  description: `Who built ${APP_NAME}, why it exists, and what it isn't. Independent, MIT-licensed, no affiliation with any AI agent vendor.`,
+  description: `Who built ${APP_NAME}, why, and what it is not. Independent, MIT-licensed, and not tied to any AI agent company.`,
 };
 
 const ABOUT_JSON_LD = {
@@ -59,50 +59,47 @@ export default function AboutPage() {
 
       <section className="my-3 mb-7">
         <h1 className="mb-2.5 text-[30px] font-bold leading-[1.18] tracking-tight">About</h1>
-        <p className="m-0 max-w-[72ch] text-[15.5px] text-ink-dim">
-          Who built {APP_NAME}, why it exists, and what it deliberately isn&apos;t.
-        </p>
+        <p className="m-0 max-w-[72ch] text-[15.5px] text-ink-dim">Who built {APP_NAME}, why, and what it is not.</p>
       </section>
 
       <Panel>
         <PanelHeading>Who</PanelHeading>
         <p className="m-0 text-[14.5px] leading-relaxed text-ink-dim">
-          Built and maintained by <ExternalLink href="https://github.com/hsnice16">Himanshu Singh</ExternalLink>.
-          Independent project — no affiliation with Anthropic, OpenAI, Google, Cognition, Anysphere, or any of the agent
-          vendors ranked here.
+          Built and maintained by <ExternalLink href="https://github.com/hsnice16">Himanshu Singh</ExternalLink>. It is
+          an independent project, not tied to Anthropic, OpenAI, Google, Cognition, Anysphere, or any other company
+          whose agent is ranked here.
         </p>
       </Panel>
 
       <div className="mt-3.5">
         <Panel tone="warn">
-          <PanelHeading tone="warn">Why this exists</PanelHeading>
+          <PanelHeading tone="warn">Why it exists</PanelHeading>
           <p className="m-0 text-[14.5px] leading-relaxed text-ink-dim">
-            The gap between &ldquo;repo with a README&rdquo; and &ldquo;repo that actually helps an AI coding agent ship
-            code&rdquo; keeps widening, and there&apos;s no public way to tell who&apos;s doing the work. {APP_NAME}{" "}
-            tries to make that visible — per model, because the agents aren&apos;t interchangeable. Claude Code wants an
-            AGENTS.md and a fast test loop; Cursor wants strong types and a skim-readable README; Devin wants a runnable
-            dev environment with declared deps and tests. The same repository can score very differently across them,
-            and a single overall number would hide that.
+            A repo that has a README is not the same as a repo that really helps an AI coding agent get work done. That
+            gap keeps growing, and there is no public way to see which repos have done the work. {APP_NAME} tries to
+            show it, for each agent separately, because the agents are not the same. Claude Code wants an AGENTS.md and
+            fast tests. Cursor wants strong types and a README that is easy to skim. Devin wants a dev setup it can run,
+            with its dependencies and tests listed. One repo can score very differently for each of them. A single
+            number would hide that.
           </p>
         </Panel>
       </div>
 
       <div className="mt-3.5">
         <Panel>
-          <PanelHeading>What it isn&apos;t</PanelHeading>
+          <PanelHeading>What it is not</PanelHeading>
           <p className="m-0 text-[14.5px] leading-relaxed text-ink-dim">
-            This is not a benchmark of agent performance. Today every score is derived from{" "}
-            <strong className="text-ink">static signals</strong> — file existence and content-length checks on the
-            cloned tree. No agent is actually run. Per-model rationales are derived from each agent&apos;s published
-            documentation (sources are linked on the methodology page), but the weight values themselves are still
-            pre-benchmark — not yet calibrated against measured agent success. Read the{" "}
+            It is not a test of how well agents perform. Every score comes from{" "}
+            <strong className="text-ink">simple file checks</strong>: does a file exist, and how long is it. No agent is
+            actually run. The reasons behind each agent&apos;s weights come from that agent&apos;s own docs (linked on
+            the methodology page). But the weight numbers are not yet tested against how agents really perform. Read the{" "}
             <Link
               href="/methodology"
               className="border-b border-dotted border-ink-dim/60 text-ink-dim hover:border-ink-soft hover:text-ink-soft"
             >
               methodology
             </Link>{" "}
-            for the full picture, including the production-cut plan to replace pre-benchmark weights with measured ones.
+            for the details, including the plan to replace these weights with measured ones.
           </p>
         </Panel>
       </div>
@@ -111,9 +108,9 @@ export default function AboutPage() {
         <Panel>
           <PanelHeading>Open source</PanelHeading>
           <p className="m-0 text-[14.5px] leading-relaxed text-ink-dim">
-            MIT-licensed. The signal definitions, weight profiles, scoring code, seed list, and every score in the
-            database are all in the <ExternalLink href={REPO_URL}>source repository</ExternalLink>. If a repo&apos;s
-            score looks wrong, file an issue with a link and the rubric to revisit; if a signal is missing, propose one.
+            MIT-licensed. The checks, the weights, the scoring code, the list of repos, and every score are all in the{" "}
+            <ExternalLink href={REPO_URL}>source repository</ExternalLink>. If a score looks wrong, open an issue with a
+            link and say which rule to look at again. If a check is missing, suggest one.
           </p>
         </Panel>
       </div>
@@ -123,7 +120,8 @@ export default function AboutPage() {
           <PanelHeading>Contact</PanelHeading>
 
           <p className="m-0 text-[14.5px] leading-relaxed text-ink-dim">
-            Best signal: open an issue or discussion on <ExternalLink href={`${REPO_URL}/issues`}>GitHub</ExternalLink>.
+            The best way to reach me: open an issue or discussion on{" "}
+            <ExternalLink href={`${REPO_URL}/issues`}>GitHub</ExternalLink>.
           </p>
         </Panel>
       </div>