First games admin tool. A "Games" tab in the operator console for the Daily Word
answer pool.
* Lookup: is a word real (in the guess dictionary), the right length (5/6), and
already in the pool — instant as you type.
* Add: appends to the pool, enforcing the invariant (alpha · 5/6 letters · in the
guess dict) so the daily answer is always guessable. Remove: drops admin-added
words (curated static ones stay).
* Additions persist in a new word_pool table (survives redeploys, unlike the
baked-in JSON); the daily picker reads static pool ∪ DB additions. Guess dicts
shipped with the package (goodnews/data/words-5/6.json) for server-side
validation. Admin-gated endpoints + tests.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex audit. test_wordsearch_endpoint now asserts the exact promise — small 6,
med 9, large 13, pairwise-disjoint. app.css comment updated to .playing-game (the
class was renamed when the focused viewport was generalized to both games).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per field feedback.
* Each day is now THREE distinct puzzles: the three sizes draw DISJOINT word
slices from a date-shuffled pool (small/med/large = 6/9/13, sum 28 unique).
Curated fallback themes expanded to 30 words each; LLM proposals accepted only
if they supply >= 28 unique words, else fall back. No more repeats across sizes.
* Word Search is now a focused game screen on mobile (same as Daily Word): body
scroll locked + footer hidden (generalized .playing-game), and the grid sizes
to the largest square that fits between the theme and the palette (container
query) — the whole puzzle is on screen, no page scroll.
* Theme placement: full "Today's theme · <name>" on the size-selection screen;
just the theme name on the puzzle itself, saving vertical space for Large.
* cosy → cozy. 🇺🇸
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two reported bugs, same root cause: the fixed-cell grid overflowed its wrapper
on Large, so (a) the last column spilled past the border and (b) the pointer→cell
math drifted across the row, recording finds "off by a letter".
* Grid now uses 1fr columns with max-width = n·32px: the board grows with the
grid and can never overflow (shrinks to fit a narrow phone instead).
* cellAt() accounts for the grid padding/border, so selection is exact edge-to-edge.
* restore() now validates each saved find against the CURRENT grid and drops any
whose cells no longer spell the word — clears stale highlights if the day's
puzzle changed.
Codex follow-ups:
* _ws_propose now requires >= large.count + 4 valid words before accepting an LLM
proposal (else falls back to a curated theme), so a thin LLM result can't
underfill Large. Added a thin-LLM fallback test.
* Cleaned Svelte warnings: removed the now-unused .gamecard.soon CSS, added an
ARIA role/label to the grid, declared gridEl with $state. Build is warning-clean.
* Added a stale-load guard in WordSearchGame.load() so rapid size switches can't
let an older request overwrite the newer selection.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
From playtesting findings:
* Pools nearly doubled (115/104 → 228/201) with calm/neutral everyday words
(claps, dance, drench, beach…), not just strictly-upbeat ones — more variety,
~7-month runway. The post-solve "why" prompt reworded to fit neutral words.
* Word Search now stores one theme + word list per day; the grid is built per
request for three SIZE tiers — Small (8×8, 6 words), Medium (11×11, 9),
Large (14×14, 13). Large packs more words = a longer sit ("too fast" fix).
All sizes share the day's theme; every size still code-placed + solvable.
* Word Search themes can now be neutral everyday scenes ("Around the house",
"At the beach", "In the kitchen", "A walk outdoors", "Making music"…), not
only hopeful — same shape as the articles.
* Each found word gets its own colour from a calm palette, in the grid and its
word-list chip. Per-size local progress + best time.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A calm second daily game, same philosophy as Daily Word — LLM proposes, code
disposes.
* LLM proposes a hopeful theme + ~8 words; code validates (alpha/length/dedup)
and PLACES every word in a date-seeded grid, so the puzzle is always solvable.
Curated fallback themes if the LLM is thin. Only placed words are returned;
the solution cells (placements) are never sent to the client.
* GET /api/puzzle/wordsearch → {theme, words, grid, size}. No answer to hide:
the grid and word list are meant to be seen — the play is finding them, which
the client validates by reading the selected line off the grid.
* WordSearchGame.svelte: pointer-drag selection snapped to the 8 straight
directions (mouse + touch), found-word highlighting, no-fail, no pressure
timer — time is recorded quietly and shown at the end with a personal best.
Spoiler-free share. localStorage progress (restores found cells + timer).
* Hub's Word Search card is now live with today's status; cycle pre-generates
both games with the LLM.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex's v2 hardening. The GET /api/puzzle/word response no longer carries
the answer at all — guesses POST to /api/puzzle/word/guess and the server
returns the colour pattern, computed against the day's answer. The answer (and
the "why") are revealed only once solved or the guesses are spent. This removes
the "open DevTools, read the answer" issue without pretending to be a fortress
(a deliberate crafted request can still peek; there's no leaderboard or prize,
so that's fine). Client keeps local progress/stats; dict validation stays
client-side. Trade-off accepted: each guess needs the API (the site already
depends on it for today's content).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex. Pool grown 51/44 → 115/104 hopeful answers (5/6 letter) via the
agreed workflow: LLM proposes themed candidates → code filters to the bundled
guess dictionary (length/alpha/dedup) → human spot-check prunes tone-drift
("growl", "plain", "color"…). ~3.5-month runway before repeats per variant.
test_wordpool.py locks the invariant in CI: every answer must be lowercase
alpha, correct length, unique, and present in words-5/6.json — so no future
addition can become an unguessable puzzle.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A calm /play space — "after the brief, a small thing to enjoy." Framework-ready
for more games (Word Search next; zen/coloring later).
* Daily Word (5 letters / 6 guesses) + Long Word (6 / 7) — same Wordle mechanic,
Upbeat Bytes flavor (no "Wordle" in the UI). Hopeful answers; after solving, a
one-line "why this word matters."
* LLM proposes, code disposes: answers are picked deterministically by date-seed
from a hand-curated hopeful pool that's pre-validated ⊆ the guess dictionary
(always typeable), avoiding recent repeats; the LLM only adds the optional
"why" (with fallback). daily_puzzles(date, game, variant, payload) stores them
so everyone gets the same daily; the cycle pre-generates with the "why".
* Bundled guess dictionaries (words-5/6.json, ~12.6k/22.4k) for client-side guess
validation — never the LLM. Answer lightly obfuscated (base64) in the payload.
* Private, gentle stats (played/solved/streak, guess distribution); spoiler-free
emoji-grid share. No leaderboard, no timer, no streak-loss drama.
* Play in the bottom nav (replacing Browse, still on the lane rail) + the header.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two calm returning-reader features.
Since-last-visit (Highlights companion, not a nav lane — per Codex):
* queries.feed gains a `since` filter; GET /api/since?ts= returns the count +
a few accepted/non-dup/visible articles discovered since the reader's last
visit (boundary-respecting; invalid/future ts → 0, no error).
* Home stores last_seen in localStorage (reads prev, then stamps now); on
Highlights, a gentle "Since you were last here, N new calm reads came in"
note with a "See what's new" reveal of a compact inline section. Dismissible.
No badges, no unread counts, no "missed" language.
PWA:
* Real PNG icons (192/512 + full-bleed maskable) rasterized from favicon.svg;
manifest fixed (azure theme to match the brand, PNG icons); apple-touch-icon.
* Minimal service worker: precache the app shell, always-fresh API + /a/ pages.
* Gentle, dismissible install banner (beforeinstallprompt → Install; iOS → the
Share → Add to Home Screen hint). Never nags.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: generate_summary treated why_belongs alone as a complete explanation,
but get_explanation requires all three — so a partial older row (e.g. only
why_belongs) would never top up and the page would fall back forever. Now the
fully-cached check requires summary + what_happened + why_matters + why_belongs.
Test covers the partial-row top-up.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex — make /a/<id> feel like Upbeat Bytes has editorial judgment, not just
a summary wrapper. Trust-building, short, not an essay.
* article_summaries gains what_happened / why_matters / why_belongs (+ migration).
* summarize.explain_article: a separate, fallback-able LLM pass producing three
short notes (parsed from a labelled WHAT/MATTERS/BELONGS format). generate_summary
now stores them alongside the summary, and tops up older summaries on next view.
get_explanation returns them only when all three are present.
* API: share_page + /api/summary expose the explanation.
* share.py: renders the three-part section (accent rule) when complete; otherwise
the single "Why it's here" reason line is the calm fallback. The page polls and
swaps in both the summary and the section as they cache.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex — give follows an immediate payoff without making the email feel
algorithmic. The editorial brief stays the star: subject + brief count unchanged.
* followed_digest_items(conn, user_id, exclude_ids, limit=3): recent items from
the user's followed sources/tags, same accepted/non-dup/content-visible gate,
excludes anything in the brief, capped to one per source so a single follow
can't dominate. Empty → section omitted (no empty state in email).
* build_digest gains an optional `followed` list → a small "From what you follow"
section AFTER the brief, only when there are items. Item rendering factored into
shared _item_html / _item_text_lines helpers.
* send_due_digests computes the followed items per user (excluding the brief).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex — turn accounts into a real reason to return, without an algorithmic
feed. Durable interests (sources + tags), not moods.
* DB: user_follows (user_id, kind source|tag, value, unique).
* queries.feed gains follow_sources/follow_tags → the Following feed is
"articles from a followed source OR carrying a followed tag", still respecting
calm filters/boundaries.
* API: GET/POST/DELETE /api/follows (sign-in required; source ids validated);
/api/feed?following=true resolves the user's follows (anon → empty, not error).
* Frontend: follows store (followKeys + toggleFollow, mirrors savedIds); a
Follow button on source + tag/topic views; a "Following" lane in the nav with
a tailored empty state; a Following management section in Account (unfollow).
Digest "From what you follow" deferred to v2 (brief stays first).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per user feedback: rename from 'Today's good news' (which implied the site was
done for the day) to a publication-style 'Upbeat Bytes — Daily Highlights'
masthead with a warm morning intro and a sign-off that links back to the site
('more good news is always waiting'). Adds 'why it's here' to the plain-text
part too. No images by design (lightweight, mail-client-safe).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* On-site end-cap now says "You're caught up for now." — honest, since Highlights
refreshes through the day (the email keeps the daily "see you tomorrow").
* Anonymous "Get tomorrow's brief by email" now honors the one-tap promise:
sets a pending flag, opens sign-in, and auto-enables once auth resolves.
* Email compliance (RFC 2369/8058): send_email takes optional headers; the digest
sets List-Unsubscribe + List-Unsubscribe-Post=One-Click, and a POST
/api/digest/unsubscribe handles native one-click (GET still serves the page).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex — a per-row Check button that previews a LIVE source on demand,
intentionally read-only and ephemeral.
* POST /api/admin/sources/{id}/preview — admin-gated, safe-fetch + heuristic
preview (reuses the candidate preview path), returns the result. Mutates
NOTHING: no DB write, no poll attempt, no health/state change. 404 on missing.
* UI: per-row Check button with a Checking… state; results in an inline row
under the source (sampled, would-pass %, recent-7d, example accept/skip
headlines) with dismiss; inline error on failure. "Checked just now" is
local UI state only. Heuristic v1 — model deep-check left for later.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex — make the table more decision-ready from data we already have.
Paywall is a domain-level hint, so it's a per-source flag (not a meaningful
rate): show image-coverage % plus a 🔒 marker for subscription domains in one
compact "Media" column (tooltip spells it out). source_health gains a
`paywalled` flag (is_paywalled on homepage/feed); also added to sources.csv.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: outgoing reply now sets Reply-To = GOODNEWS_REPLY_TO_EMAIL, falling
back to the From address. Never the reader's own address (they're the recipient).
send_email gained an optional reply_to param. Failed-send stays UI-only (draft
kept) — no schema change, per Codex's lean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: source-controlled strings (name, feed_url, last_error, review_reason)
could be read as formulas by spreadsheet apps if they start with = + - @. Add
_csv_cell — prefixes such strings with an apostrophe; numbers pass through
untouched (no risk, and avoids mangling negatives). Routed every exported cell
through it. Test: a =HYPERLINK(...) source name is escaped, never bare.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex v1 — boring-in-the-best-way: inspect/archive operational data outside
the app. Admin-gated, Python csv module, text/csv + attachment disposition.
* GET /api/admin/export/sources.csv — current-state snapshot per source: name,
feed/homepage, status, visible, served/accepted/total, acceptance/duplicate/
accepted-dup/image-coverage %, last success/error, retry-after, review.
* GET /api/admin/export/audience.csv?days= — summary block (visitors, returning,
accounts, feedback, shares) + a blank line + the daily visits/opens series;
range applies to audience, sources is a snapshot.
* source_health now also returns feed_url/homepage. Small download links on the
Sources + Audience tabs.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: a paused/retired source with a future retry_after_at shouldn't nag
'rate-limited for 12h+' — it's intentionally out of polling. Scope long_rest to
active (matching the other operational items). Test: paused/retired rate-limited
sources stay quiet.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex — make the Overview strip diagnostic without making the operator hunt
through tables. Aggregated (one calm line per condition with a count), volume-
gated, conservative thresholds:
* Stale: active+visible source, last success > 10 days ago (warn).
* High rejection: >=20 ingested, acceptance < 25% (info).
* High duplicate: >=10 accepted, accepted-dup > 50% (info).
* Thin images: >=10 served, per-source image coverage < 25% (info).
* Long rate-limit: retry_after_at more than 12h out (info).
source_health gains a per-source images count + image_coverage. _attention takes
an optional now (for tests). Existing site-wide items (global image coverage,
thin brief, unread feedback) unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: the Next poll column computed only the streak-backoff time, so a
rate-limited source could show an earlier Next poll than the real gate (which
also requires retry_after_at <= now). Take the later of the two in the Python
post-process so the admin table agrees with due_source_rows.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex's spec — a publisher saying "slow down" shouldn't make a feed look
broken, but repeated 429s stay visible via last_success_at / stale-source.
* Schema: sources.retry_after_at (nullable) + migration.
* feeds.parse_retry_after: delta-seconds OR HTTP-date → UTC stamp; ignores
invalid/negative/past; caps at now + MAX_BACKOFF_MINUTES.
* fetch_feed raises RateLimited (carrying the parsed time) on a 429.
* poll_source: on 429 set retry_after_at + last_error, status='rate_limited',
and do NOT increment consecutive_failures; on success clear retry_after_at;
non-429 failures unchanged.
* due_source_rows requires BOTH the streak backoff elapsed AND retry_after_at
passed (i.e. the later of the two).
* Admin: source_health returns retry_after_at; status reads
"rate-limited · rests until …" rather than "failed/resting".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bring the supervised source-candidate flow into Sources (Codex's v1 scope), so
adding feeds no longer needs the CLI.
* feeds.safe_fetch_feed: SSRF-safe fetch for UNTRUSTED (admin-pasted) URLs —
http(s) only, every redirect hop re-validated via enrich._host_is_public,
body size-capped, bounded redirects, no cookies. preview_feed gains a
`fetcher` param; the API path passes safe_fetch_feed (NOT the raw fetch_feed
used for already-vetted polling).
* API (admin-gated): GET /candidates; POST /candidates (suggest+preview, gated
before the outbound fetch, no DB conn held during network); /{id}/preview
(explicit re-preview); /{id}/promote (paused by default, returns the new
source + updated candidate); /{id}/reject. rejected stays on candidates only.
* Admin Sources tab: "Add a source" field + a candidate queue showing the
preview (pass rate, recent count, example headlines) with Promote (as paused,
or Activate immediately) / Re-preview / Reject.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: upsert_sources() wrote `active` but not `status`, so a candidate
promoted inactive (the pipeline default) became active=0 + status='active' —
the exact mirror drift Phase 1 set out to avoid (scheduler won't poll, admin UI
shows "active"). Now derive status from an explicit value or from active, mirror
active off status, and write both columns together (insert + conflict update).
Test: promote_candidate(active=False) → status='paused', active=0.
Also fix stale source_health docstring (now includes retired).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex's plan — introduce a lifecycle without a risky "change the source of
truth everywhere" moment.
* Schema: sources.status (active|paused|retired) + content_visible; migration
backfills status from active (active=1→active, else paused), content_visible=1.
* `active` is kept as a SYNCED MIRROR: status active→active=1, paused/retired→0,
so the scheduler/CLI/legacy code keep working unchanged.
* Retire stops polling but keeps articles visible (non-destructive). Hiding is a
separate, reversible lever: content_visible=0 drops a source's articles from
the public feed + brief (read AND build), behind a confirm. Personal saved/
history are untouched.
* API: /sources/{id}/status (validates, mirrors active) + /visibility, replacing
/active. source_health returns status + content_visible.
* Admin: status column (active/paused/retired + "hidden"), Retired filter,
Pause/Resume · Retire/Restore · Hide/Show actions.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: slicing the SANITIZED html with [:8000] could cut through a tag or
entity. Cap the RAW editor HTML (20k) before sanitizing instead, and have
sanitize_reply_html auto-close any still-open allowed tags so malformed input
can never leave a dangling/severed tag in message_html or the email body.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the Markdown composer with a small contenteditable WYSIWYG (Codex
greenlit for this narrow, admin-only surface).
* markup.py: render_reply_html → sanitize_reply_html + reply_html_to_text.
Allowlist rebuild via stdlib HTMLParser — keeps strong/em/p/br/ul/ol/li and
span ONLY with a whitelisted font-size (13/15/18/22px); normalizes b→strong,
i→em, div→p, <font size> → safe span; drops links/images/arbitrary styles
(content kept as escaped text) and discards script/style content entirely.
* API: FeedbackReplyBody.html (raw editor HTML); endpoint sanitizes → message_html,
derives plain text → stored message + the email text/plain part. Unchanged:
multipart send, store-on-success, conn released during SMTP, mark-read, 404/400/422.
* Frontend: contenteditable editor + toolbar (Bold/Italic/Size/• List/1. List),
execCommand with styleWithCSS=false for semantic tags, font size wraps the
selection in a fixed-px span, paste intercepted as plain text. No links yet.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex's optional note: alongside the ingest-wide duplicate_rate, expose
accepted_dup_rate — of what a source got ACCEPTED, how much was a duplicate of
already-served content (accepted_total − served). Nearly free (derived from
existing counts); surfaced as a tooltip on the Dup column so the table stays
uncluttered.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per Codex: a constrained Markdown-ish composer rather than contenteditable.
* goodnews/markup.render_reply_html — escapes everything first, then introduces
only a tiny whitelist (**bold**, - bullets, #/##/### headings, paragraphs,
line breaks). No links, attributes, inline styles, or raw HTML passthrough.
* feedback_replies.message_html column (+ live migration); replies store both
the Markdown text and the rendered HTML.
* email_send.send_feedback_reply now sends multipart text/plain + text/html
(the sanitized render, wrapped in a trusted email template).
* Frontend: textarea + a small toolbar (Bold / • List / H) that inserts
Markdown; the reply thread renders the server-sanitized HTML.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reply to a reader from the admin inbox instead of a mailto. Per Codex: keep v1
plain text (no rich editor — defers the user's bold/bullets ask as a fast-follow).
* DB: feedback_replies table (feedback_id, user_id, message, sent_to, sent_at),
created on the live DB.
* email_send.send_feedback_reply: plain-text "Re: Your Upbeat Bytes feedback"
with a quoted context block, no analytics/account details.
* API: POST /api/admin/feedback/{id}/reply — admin-gated, requires the feedback
exists (404) and has a contact_email (400), trims+caps the message; sends via
SMTP and only records the reply on success (502 on send failure so the UI keeps
the draft); marks the item read. Feedback list now includes each item's replies.
* Frontend: inline composer (Send/Cancel, sending state, error keeps draft) +
reply thread under the message; Reply only shows when there's an address,
else "No reply address".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
/api/admin/stats accepts ?days= (clamped to 7/30/90, default 30) → passed to
admin_stats, which already windows visitors, retention, funnel, sharing, daily
trend, and the top lists by that span. Frontend adds a Window picker on the
analytics tabs (Overview/Content/Audience); changing it refetches and the
windowed labels ("Visitors (Nd)", "Last N days", "Returning visitors (Nd)")
follow. Corpus totals + source health are unaffected (not time-windowed).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Turn the Sources tab into a real management console (per Codex):
* source_health now lists ALL sources (active + paused) with backing metrics:
served / accepted_total / total_articles / duplicates + acceptance & duplicate
rates + review_reason, alongside last success/attempt, next poll, failures.
* Admin endpoints (gated, 404 on missing): POST sources/{id}/active (pause/
resume) and /review (flag/clear with reason).
* Pausing only stops future polling — the feed query has no active filter, so a
paused source's accepted articles stay live.
* Frontend: metric table + Paused filter + per-row Pause/Resume & Flag/Clear
(optimistic, revert on failure). Attention 'resting' now scoped to active.
Retire/Delete intentionally deferred (distinct lifecycle state, later).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per audit: read-toggle and delete returned {"ok":true} even for a nonexistent
id. Return 404 when no row is affected, so the optimistic UI can distinguish a
stale/already-deleted row from a real success. (The postJSON/delJSON imports
flagged in the audit were already present — verified in source + built bundle.)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make the admin Feedback section a real inbox.
* DB: feedback.read_at column (schema + idempotent migration).
* API: feedback list returns read_at; POST /api/admin/feedback/{id}/read
{read} toggles it; DELETE /api/admin/feedback/{id} removes a message
(both admin-gated). admin_stats gains feedback_unread; the Attention strip
and the tab badge now count UNREAD, not total.
* Frontend: unread messages are highlighted with an accent rail + dot; an
Unread filter joins the category chips; each message has Mark read/unread
and Delete (confirm), with optimistic updates that revert on failure.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Click a source name on any card → a feed of just that source's articles,
newest-first, still accepted / non-duplicate / boundary-filtered (the calm
promise isn't bypassed). A natural way to follow a publication's feel.
* queries.feed + /api/feed: source_id filter; Article output gains source_id.
* Frontend: source label is a button → transient 'source:<id>' view (like
'tag:<slug>'), rendered in the feed grid with Load more, header = source name.
* Ad-hoc, not a pinned lane. Foundation for a future source page (metadata) +
Follow; shareable /source/<slug> route and source_view analytics come then.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make "no blurry images" sustainable, not a one-off cleanup. RSS feed thumbnails
(~44% were ~90px) were stored at ingest and upscaled to mush, so new articles
would reintroduce them. Now image_url is filled ONLY by the quality-gated
og:image enrichment:
* insert_article no longer stores the feed image (was canonicalize_url(item...)).
* enrich_recent_images(): the cycle fetches a quality og:image for the newest
accepted, imageless articles each run (bounded), keeping Latest photo-rich.
* Brief + on-open enrichment unchanged.
Net: every stored image is a validated, ≥450px og:image; the rest are clean
placeholders.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Some cards showed blurry photos — feed RSS thumbnails (~90×90, e.g. Phys.org's
/tmb/ path) that load fine but upscale to mush in the banner. Add a header-based
dimension parser (PNG/GIF/JPEG/WebP, stdlib only) and fold a minimum-size gate
(450×250) into the image validation, alongside the existing load check. Images
we can't measure (SVG/AVIF) still pass on content-type. A re-prune clears the
small ones already stored so those cards fall back to the clean placeholder.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A stored og:image isn't proof it renders: signed/hotlink-protected URLs (e.g.
the Guardian's i.guim.co.uk) 401 on a direct browser load, so they counted
toward coverage yet always fell back. Now fetch_og_image confirms the image
truly returns 200 + image/* (requested no-referrer, same SSRF-safe redirect
handling) before storing it. Add prune_broken_images() to clear already-stored
URLs that no longer load, so coverage is honest and those cards show the
placeholder cleanly. The browser onerror→placeholder remains the final safety net.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Tie image enrichment to attention (per review): when an article earns a summary
(i.e. a reader reached it), best-effort fetch a real og:image if it lacks one —
never blanket-fetch every ingested article. Adds:
* enrich_article_image() — single-article fetch, leaves existing images alone,
retries an imageless article only after 7 days, stamps image_checked_at.
* generate_summary() calls it after caching (wrapped; never breaks summaries).
* enrich_summarized_images() + `goodnews enrich-images` CLI — slow background
backfill of already-summarized, accepted, imageless articles.
* Quality gate: extend the generic-image skip list with data:/tracking-pixel/
spacer markers (on top of the existing logo/placeholder + unbranded-BBC logic).
This is coverage only; display (editorial rhythm, tile treatment) comes next.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Restructure the nav around two permanent lanes, then the reader's chosen ones:
"Highlights" (the curated daily brief — formerly "Today") and "Latest" (the
freshest accepted stories, newest-first). Now that the gate is tight, a
chronological "incoming" feed is safe to expose.
* feed(): new sort="latest" (pure recency) alongside the default best-first
rank; /api/feed exposes sort=ranked|latest (validated). Still accepted-only
and boundary-respecting either way.
* lanes.py: two pinned lanes (Highlights + Latest) instead of one.
* Home: "Latest" view + "Load more" pagination for every feed view (offset-
paged, de-duped). Mobile bottom bar gains a Latest tab.
* LanePicker shows both pinned lanes; nav rail renders them first.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Date fix: introduce GOODNEWS_TZ (goodnews/localtime.py) so the brief's "today"
rolls over in a pinned zone (Eastern) instead of UTC — robust to host-clock
resets. The home page now formats the brief's date in each VISITOR's local
timezone (from its UTC freshness stamp), so nobody ever sees "tomorrow."
* Admin "Content served": articles live, fresh (7d), ingested (24h), summaries,
active sources, today's brief size — queries.content_stats().
* Admin "Source health": per active source, the failure streak, last error,
accepted contribution, and computed next-poll time (so backoff / "resting
until" is visible), via queries.source_health() reusing the feeds backoff
math. Failing sources sort to the top; times render in the viewer's zone.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Pre-traffic cleanup from an audit:
* Scheduler: poll_due_sources now keys on the last *attempt* (success or
failure), not the last success, and scales the wait by the consecutive-
failure streak (capped at a day). A failing feed (e.g. Phys.org's HTTP 429s)
used to be retried every cycle because it had no successful run; it now backs
off and recovers on its own. Extracted due_source_rows() + tests.
* FK hygiene: deleting a daily_brief is supposed to cascade to its items, but
SQLite enforces foreign keys per-connection — connect() already sets the
pragma, so the cascade is correct going forward; added a regression test.
(Orphaned items + Phys.org settings were cleaned directly on the live DB.)
* a11y: modal/drawer dialogs are now focusable (tabindex), close on Escape
(window) and on backdrop click via a target check (dropping the inner
stopPropagation handlers). Build is warning-free.
* tests: conftest points any un-mocked LLM client at a closed port with a 1s
timeout, so an accidental real call fails fast instead of hanging the suite.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Readers can now choose which quick-access lanes sit above the feed; "Today"
stays pinned. The pool (goodnews/lanes.py, served at /api/lanes) is one source
of truth over three lane kinds the feed already renders: moods, primary topics,
and high-volume Discovery tags. Selection lives in the existing prefs blob
(localStorage + /api/prefs sync); the filter parser ignores the new `lanes`
field, so it rides along harmlessly. Default = today's moods, unchanged.
Food/Space stay grouping tags rather than primary topics (per review): `space`
already existed; added `food` to the Mind & Craft family so the classifier
assigns it, and seeded the Food lane by re-tagging the two food sources.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make the summary pages discoverable so traffic compounds passively:
- /today: a server-rendered, shareable + indexable digest of today's brief —
each item's title (→ /a summary), our summary, and a source link. OG/Twitter
meta + self-canonical.
- /sitemap.xml: dynamic — home, /today, and every accepted non-duplicate /a page
with lastmod. robots.txt allows all and points to it.
- Home (SPA shell) gains canonical + OG/Twitter tags for cleaner unfurls.
- Caddy routes /today + /sitemap.xml to the API. 133 tests pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Feedback:
- feedback table; POST /api/feedback (anonymous-ok, optional category/email,
honeypot + per-day flood cap) stores + emails the admin; GET /api/admin/feedback.
- Shared feedback store + FeedbackModal; a speech-bubble opens it from the desktop
header, the mobile top bar (logo moves left), the footer, and /account. Feedback
section in /admin.
Stats (additive, same privacy model — no IP/UA/referrer/raw terms):
- Event vocab: summary_viewed (fired on /a load), full_story (card → source),
not_today/less_like_this/hide_topic, replace_used/replace_none, paywall_replace,
paywalled_source_open. Card title/image opens /a (no double-count); history
records via keepalive so it survives the nav.
- Dashboard: Accounts card (counts only), reading funnel (summary→source rate),
emotional-mix & friction, paywall, returning-visitor buckets. (Health metrics
deferred to a future monitoring dashboard.) 131 tests pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
thejayman77