Değişiklik günlüğü

GitHub

Her önemli değişiklik, sürüm sürüm. Aynı dosya depoda yer alır ve 19 dilin tümüne çevrilmiştir.

[1.46.0] — 2026-05-18

WS2 #5 + #6 + #21 + #24 — scan SSE accessibility.

🐛 Fixes

fix(a11y): scan SSE — live-log region, Stop, run-state, error banner — four UX-audit findings on #/scan:
- #5 the streaming console (<pre id="scan-console">) was a bare element; screen readers got no announcement of scanned lines. Now role="log" aria-live="polite" aria-relevant="additions" + an aria-label + tabindex="0" (keyboard-scrollable). A separate visually-hidden role="status" aria-live="assertive" region announces terminal events (complete / failed / stopped).
- #6 an in-flight EventSource scan had no abort path. The handle is now captured (activeES) and a Stop button closes it (es.close()), cancels the result poll, and resets state. Stop is shown only while a scan runs.
- #21 the Scan button stayed enabled with no busy cue during a multi-minute crawl. setScanRunning() now disables it + sets aria-busy and toggles the Stop button across both stream paths (single-phase streamTo and multi-phase runScanAll — the latter only ends the run on the terminal done, final !== false).
- #24 an SSE failure was a 3.5 s toast only. A persistent role="alert" banner now shows the error with a Retry scan action (re-invokes the last run fn); cleared on the next run.
All side-effecting closures keep the existing __cancelActiveScanPoll() hashchange cleanup.

🌐 i18n

8 new keys × 8 locales — scan.consoleLabel/stop/stopped/statusDone/statusFailed/statusStopped/errBannerTitle/errRetry. i18n-coverage gate green.

🧪 Tests

test: tests/scan-sse-a11y.test.mjs — 7 cases (log-region roles, assertive status, Stop closes EventSource on both paths, run-state aria-busy, persistent alert + Retry, terminal-done gating, 8 i18n keys ×8). 660 → 667.

Verification

$ npm run test:ci
# 667 / 667 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.46.0
# Playwright #/scan: console role=log aria-live=polite tabindex=0 · status role=status · error role=alert(hidden) · Stop hidden · 0 errors

[1.45.0] — 2026-05-18

WS2 #3 — config tabs: full WAI-ARIA Tabs pattern.

🐛 Fixes

fix(a11y): config.js tabs implement role=tablist/tab/tabpanel — the three #/config tabs (API keys / Profile / Modes) were plain <button class="tab-btn"> with click-only activation: no role, no aria-selected, no keyboard model (UX-audit HIGH #3, WCAG 4.1.2 / 2.1.1). Now: a role="tablist" container with an aria-label; each tab role="tab" + id + aria-controls + aria-selected (synced in activate()) + roving tabindex (0 active / -1 rest); the panel role="tabpanel" + tabindex="0" + aria-labelledby tracking the active tab. Full keyboard nav: ←/→/↑/↓ (wrapping) + Home/End move focus AND activate. The legacy .tab-btn.is-active CSS hook is preserved. Verified live: ArrowRight API→Profile syncs aria-selected + panel labelledby; End→Modes; 0 console errors.

🌐 i18n

1 new key × 8 locales — config.tablistLabel (“Settings sections”). i18n-coverage gate green.

🧪 Tests

test: tests/config-tabs-aria.test.mjs — 7 cases (tablist/tab/tabpanel roles, aria-controls + roving tabindex, activate() aria-selected sync, keyboard map incl. wrap + preventDefault + focus move, legacy textContent-toggle removed, i18n key ×8). Total 653 → 660.
fix(test): retarget 2 stale auto-pipeline smoke tests (commit 5d253ba) — the pre-v1.34 Playwright-e2e smoke tests asserted a transient modal the dashboard “Auto-pipeline” button stopped opening in v1.34.0 (→ Router.go('/auto')); they had been red on the separate Playwright-e2e CI job for 10 releases. Retargeted to the #/auto screen. Local smoke 16/16; CI green.

Verification

$ npm run test:ci
# 660 / 660 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.45.0
# Playwright: #/config ArrowRight/End move + activate tabs, aria-selected synced, 0 errors

[1.44.0] — 2026-05-18

WS2 #4 + #9 — focus-trapped confirmation for destructive parent writes.

🐛 Fixes

fix(a11y/safety): UI.confirm() gate before whole-file parent overwrites — two UX-audit HIGHs, both data-loss: (#4) config.js saveProfileRaw/saveModesRaw replaced the ENTIRE parent config/profile.yml / modes/_profile.md with no confirmation; (#9) tracker.js Normalize/Dedup/Merge rewrote parent data/applications.md in place with no confirmation. Added UI.confirm(title, message, opts) to public/js/api.js — a focus-trapped dialog reusing the existing WAI-ARIA modal infra (focus-return, Tab-trap), returning Promise<boolean>. A new _onClose hook fires from closeModal() so EVERY dismissal path (Esc / backdrop / × / Cancel) resolves false; only the explicit confirm button resolves true. Focus defaults to Cancel (safe choice for a destructive op). NOT native confirm() (auto-dismissed in embeds, not focus-trapped). All three call sites are now gated before the API.put/API.post. Verified live: Normalize → focus-trapped “Переписать applications.md?” modal, focus on Cancel, Cancel dismisses with no POST, 0 console errors.

🌐 i18n

8 new keys × 8 locales — common.confirm, config.rawConfirmTitle/Ok, config.profileRawConfirmBody, config.modesRawConfirmBody, track.fixConfirmTitle/Body/Ok. The {op} placeholder in track.fixConfirmBody is preserved verbatim across all locales (runtime-substituted). i18n-coverage gate green.

🧪 Tests

test: tests/confirm-gate.test.mjs — 8 cases: UI.confirm exported, _onClose resolves false on every dismissal path, Cancel-default focus, modal() back-compat onClose arg, all 3 destructive sites gated before their write, no native confirm() left, 8 i18n keys present with {op} intact. 644 → 652.

Verification

$ npm run test:ci
# 652 / 652 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.44.0
# Playwright: #/tracker Normalize → focus-trapped modal, focus=Cancel, Cancel→no POST, 0 errors

[1.43.0] — 2026-05-18

User-requested — career-ops-ui open + autostart browser-raise.

✨ Features

feat(cli): career-ops-ui open — open AND raise the dashboard tab — after setup/run, bare open/xdg-open left the dashboard tab in the background when the browser was already running, so the user had to hunt for it. New scripts/open-dashboard.mjs builds the URL from HOST/PORT (rewriting a 0.0.0.0 bind to loopback), optionally waits for /api/health, opens the default browser, then force-raises it — macOS osascript activating whichever of Chrome/Brave/Edge/Safari/Arc/Firefox is running, Linux xdg-open+wmctrl, Windows start. Exposed as the career-ops-ui open verb (aliases dash, focus). bin/start.sh autostart now delegates to it so the tab is raised automatically; NO_OPEN=1 disables auto-open for headless/CI starts. Verified live: career-ops-ui open → URL printed, browser raised, exit 0.

🧪 Tests

test: tests/open-dashboard.test.mjs — 8 cases: dashboardUrl (defaults / PORT / 0.0.0.0→loopback / explicit HOST), openAndRaise platform routing (darwin/win32/linux, no real browser), waitForHealth bounded-timeout against a dead port, and static guarantees that the dispatcher routes open|dash|focus and start.sh delegates + honors NO_OPEN (old bare-open path gone). 636 → 644.

📝 Documentation

README ×8 + help-bundle §1 ×8 — the open verb added to the launch block + a note that setup/run now raises the tab automatically and NO_OPEN=1 disables it. H2-section parity preserved (17).

Verification

$ npm run test:ci
# 644 / 644 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.43.0
$ career-ops-ui open --no-wait    # URL printed, browser raised, exit 0

[1.42.0] — 2026-05-18

WS2 fix #2 — #/portals dead-route → config deep-link.

🐛 Fixes

fix(router): #/portals 404 → alias to config + Regional-sources deep-link — #/portals was an unregistered route that rendered the 404 view, even though it is a plausible bookmarked/typed URL for portal-source management (UX-audit HIGH #2). Added portals: 'config' to router.js ALIASES (same bookmark-stability pattern as settings→profile), so it now resolves to the config view with the config nav item active. When a Regional-sources group exists, the view (config.js) detects the #/portals hash, force-opens that <details> group, scrolls it into view, and moves focus to its summary (overriding the default h1 focus) so the user lands exactly on the portal-source controls. Never renders an empty regional group from the alias alone. Verified live: #/portals → config view, is404:false, active nav = config, 0 console errors.

🧪 Tests

test(router): portals→config alias guarantee — router.test.mjs static assertion guarding the new ALIASES entry against future-refactor regression. 635 → 636.

📝 Documentation

help-bundle §5 × 8 — a “Shortcut” blockquote noting #/portals now resolves to App settings / Regional sources instead of 404ing. H2-section parity preserved (17 each).

Verification

$ npm test
# 636 / 636
# Playwright: #/portals → {h1:"App settings", is404:false, activeNav:"config"} · 0 console errors

[1.41.0] — 2026-05-18

WS2 — senior UX/usability audit + cross-cutting focus-management fix.

A 10+ yr heuristic audit (Nielsen × WCAG 2.2 AA × project conventions) of all 17 routes produced a 40-finding, severity-ranked queue (.planning/.../UX-AUDIT.md); HIGH→MEDIUM→LOW are now shipped one fix per release. This release lands the #1 cross-cutting HIGH.

🐛 Fixes

fix(a11y): move focus to the new view on every route change — router.js render() replaced #content on each hashchange but never moved focus, so keyboard / screen-reader users stayed on the destroyed node and lost their place (WCAG 2.4.3 Focus Order / 4.1.3 Status Messages — cross-cutting, affected all 17 screens). New focusNewView(content) focuses the new view’s first h1/.page-title (concise SR announcement + correct focus order), making the heading focusable (tabindex=-1) if needed and falling back to #content. The very first paint is skipped so it never fights the skip-link. Wired on both the success and error render paths. Verified live: after nav, document.activeElement is the new view’s H1.page-title.

🧪 Tests

test(router): focus-management static guarantees — 4 cases in router.test.mjs (helper defined, heading-target + content fallback, first-paint skip guard, ≥2 call sites). 631 → 635.

📝 Documentation

.planning/.../UX-AUDIT.md — full 40-finding audit + prioritized fix queue + per-release ship grouping (v1.42 → v1.51). Drives the remaining WS2 iterations.

Verification

$ npm run test:ci
# 635 / 635 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.41.0
# Playwright: #/dashboard → #/config → #/tracker · activeElement = new H1.page-title · 0 console errors

[1.40.0] — 2026-05-18

WS8.3 — docs-actualization sweep + career-ops-ui help fix + askSecret hardening.

🐛 Fixes

fix(cli): career-ops-ui help no longer leaks shell source — the dispatcher printed its header comment with sed -n '2,12p', but line 12 (set -euo pipefail) is code, not a comment, so career-ops-ui help (and the unknown-verb usage text) ended with a stray set -euo pipefail line. Narrowed to 2,11p (the comment block) in both the help and *) cases. help exits 0, unknown verb exits 2 — verified.
fix(cli): scripts/init.mjs key entry never echoes — the v1.39.0 follow-up replaced the cosmetic readline-overwrite mask with a real raw-mode reader: setRawMode(true) + a buffered line so typed/pasted key bytes never reach the terminal at all (no scrollback / tmux / screen-share leak). A full VT escape FSM consumes every CSI/SS3/OSC/DCS/SOS/PM/APC sequence so arrow & function keys can’t corrupt the secret; stdin is dependency-injected so the non-TTY fallback is unit-tested without poking the global. Iterated to a clean AI-review LGTM.

📝 Documentation

README ×8 — the old “one-command install” section is replaced by a prominent “Launch & initialize in one command” section: the curl one-liner plus the explicit career-ops-ui CLI chain (clone → npm link → setup → init → doctor → run → help), the provider-wizard explanation, the CI --provider --anthropic-key --yes form, and the LLM_PROVIDER note. All 8 README badges actualized from the stale v1.22–v1.24 / tests-461–474 to v1.40.0 / tests-631 (e2e badge made non-numeric to avoid an invented count).
help-bundle ×8 §1 — a “One-command launch & init” blockquote callout added at the top of the Quick-start playbook (before “A. Setup”) in all 8 locales. H2-section parity preserved (17 each — CI gate green).

🧪 Tests

test(init): non-TTY askSecret fallback — provider-selector.test.mjs gains a DI-stdin case asserting askSecret delegates to plain ask() (trim-parity) off a TTY without mutating the shared global. 629 → 631.

Verification

$ npm run test:ci
# 631 / 631 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.40.0
$ career-ops-ui help     # clean, exit 0 (no `set -euo pipefail` leak)
$ career-ops-ui doctor   # all required green, exit 0

[1.39.0] — 2026-05-18

WS8.2 — LLM provider selector + OpenAI/Codex key + interactive init wizard.

✨ Features

feat(config): LLM_PROVIDER selector + OPENAI_API_KEY — server/lib/env-config.mjs adds LLM_PROVIDER (auto|claude|gemini) + OPENAI_API_KEY to KNOWN_KEYS/KEY_GROUPS; OPENAI_API_KEY is secret-masked. New providerOrder(env) helper: auto→[anthropic,gemini] (legacy), claude→[anthropic], gemini→[gemini]. All 6 provider-gate sites in server/lib/routes/llm.mjs (evaluate/deep/mode × Anthropic/Gemini) consult it via _provGate() — a forced provider with no key falls through to the manual-prompt path exactly like the pre-v1.39 no-key behaviour (zero behaviour change for auto/default). #/config API-keys tab gains an LLM_PROVIDER select + OPENAI_API_KEY input. Provider-key set matches what santifer/career-ops actually implements (Gemini = parent gemini-eval, Anthropic = web-ui SDK + Claude Code, OpenAI = Codex/OpenCode CLI side); Mistral/Qwen are model names with no parent env key — not invented.
feat(cli): interactive career-ops-ui init — scripts/init.mjs is now a real wizard (was a WS8.1 stub): pick provider 1-4, enter key(s), writes parent .env via the validated env-config.updateEnvFile path (explicit user action, same as POST /api/config). Flag-driven too: --provider --anthropic-key --gemini-key --openai-key --yes.

📝 Documentation

help-bundle §2 × 8 + docs/sdd/CONVENTIONS.md — provider selector + the 3 parent-implemented provider keys. (Full README ×8 + canonical-docs fold = WS8.3/WS10, the user-mandated final steps.)

🧪 Tests

test: tests/provider-selector.test.mjs — 7 cases: providerOrder (auto/unknown/whitespace → legacy, claude/gemini forced), env-config surface (KNOWN/SECRET/LLM_PROVIDERS), init parseArgs+buildUpdates (clamp, non-empty-only), and a static canary that all 6 llm.mjs gate-sites use _provGate(). 622 → 629.

Verification

$ npm run test:ci
# 629 / 629 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.39.0
$ career-ops-ui init --provider gemini --gemini-key … --yes   # writes parent .env
$ career-ops-ui doctor                                          # verifies

[1.38.0] — 2026-05-17

WS8.1 — unified CLI dispatcher + doctor verb.

AutoResearchClaw-style one-command workflow. bin/career-ops-ui.sh dispatches setup / run / doctor / init / help; package.json bin.career-ops-ui points at it.

✨ Features

feat(cli): bin/career-ops-ui.sh dispatcher — setup → bin/setup.sh (existing one-command bootstrap), run → bin/start.sh, doctor → scripts/doctor.mjs, init → scripts/init.mjs (WS8.1 stub; interactive provider wizard = WS8.2), help. Backward-compat career-ops-ui-start bin alias kept.
feat(cli): scripts/doctor.mjs — standalone health check that reuses the exact /api/health engine (spins createApp() in-process on an ephemeral port → renders the JSON to a colorized terminal report). Single source of truth — doctor can never drift from the Health page. Exit 0 iff every REQUIRED check is green, exit 1 otherwise, so setup / CI can gate on it. No new deps; read-only.

📝 Documentation

docs/sdd/CONVENTIONS.md — “CLI dispatcher” section. help-bundle §1 × 8 — CLI quickstart note. (Full README ×8 quickstart block lands in WS8.3, the user-requested final verification step.)

🧪 Tests

test(cli): tests/cli-doctor.test.mjs — 6 cases: formatReport pure logic (all-pass / required-fail / optional-only-fail / empty-tolerant), dispatcher verb-routing canary, package.json bin wiring. 616 → 622.

Verification

$ npm run test:ci
# 622 / 622 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.38.0
$ node scripts/doctor.mjs   # ✓ all required checks pass · exit 0

[1.37.0] — 2026-05-17

WS7 — pre-commit AI review in the git workflow.

Per user request: an AI review before every commit. Two layers — a fast deterministic floor that fails-HARD, and an advisory AI pass that fails-SOFT.

✨ Features

feat(workflow): pre-commit AI review
- scripts/ai-precommit-review.mjs — deterministic floor (fail-HARD): rejects a commit that stages a .env/secret-bearing file, contains a high-confidence secret pattern in the added diff lines (.env.example placeholders exempt), leaves .also( in a staged view (mirrors the CI gate), or stages a .mjs/.js that fails node --check. AI layer (fail-SOFT): runs claude -p over the staged diff when the CLI is on PATH and AI_REVIEW !== 'off'; missing CLI / offline / timeout → notice, never blocks.
- .githooks/pre-commit + scripts/install-hooks.mjs — npm install wires core.hooksPath=.githooks via the new prepare script. Idempotent; no-op outside a git checkout.
- AI_REVIEW=off git commit … skips only the AI layer. Never --no-verify (CLAUDE.md hard rule #7); CI runs the full gate regardless.

📝 Documentation

docs/sdd/CONVENTIONS.md — new “Pre-commit AI review” section (floor vs AI layer, env switch, never-bypass rule).

🧪 Tests

test(workflow): tests/ai-precommit-review.test.mjs — 6 cases over the exported pure floor functions: blocked-path detection (.env.example exempt), secret-pattern hits (added-lines only, placeholders ignored), .also( view leftover, aggregate floor, clean-diff pass. 610 → 616.

Verification

$ npm run test:ci
# 616 / 616 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.37.0
# This very commit was gated by the new hook (live dogfood).

[1.36.0] — 2026-05-17

WS6.3 — #/config Modes tab: raw blob → per-section editor. WS6 complete.

modes/_profile.md is a prompt-engineering doc (markdown tables + prose), not key→value settings — so section-level editing is the right granularity (not field decomposition). This finishes WS6: every settings surface is now structured.

✨ Features

feat(config): per-section _profile.md editor
- server/lib/routes/content.mjs — byte-exact splitProfileSections (preamble + per-## { heading, headingLine, body } via string slices, not line-join — round-trip is byte-identical) + joinProfileSections. GET /api/modes/_profile now also returns { preamble, sections }. PUT accepts { sections: { "<heading>": "<body>" } }: replaces only named sections, preamble + unknown sections + ordering survive byte-for-byte (merge-not-replace). Unknown heading → 400. Existing stripDangerousMarkdown sanitization retained. Legacy { markdown } raw path unchanged.
- public/js/views/config.js — one collapsible textarea per ## section (label = ## heading); Save sends { sections }. Collapsed Advanced: raw markdown disclosure retained for add/remove-section + preamble edits.
- i18n: 5 keys × 8 (config.modesSectionHint/modesNoSections/modesRawToggle/modesRawHint/modesRawSave).

📝 Documentation

help-bundle §2 × 8 locales — Modes tab documented as per-section editor (merge-by-section, byte-exact preservation, Advanced raw disclosure).

🧪 Tests

test(config): tests/modes-section-form.test.mjs — 6 cases: GET exposes preamble+sections, section-merge preserves preamble/others byte-exact, unknown-heading 400, ordering preserved, legacy raw path, sanitization. 604 → 610.

WS6 outcome

Every settings surface is now field/section-structured: API-keys (already field-based — WS6.2 audit confirmed KNOWN_KEYS ≡ FIELDS, no gap), Profile scalars (WS1 v1.32.0), Profile arrays (WS6.4 v1.35.0), Modes sections (WS6.3 v1.36.0). The raw editors remain as documented escape hatches.

Verification

$ npm run test:ci
# 610 / 610 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.36.0

[1.35.0] — 2026-05-17

WS6.4 — #/config Profile structured array editors + WS6.2 API-keys audit.

WS1 (v1.32.0) gave the 14 scalar profile fields a form. WS6.4 finishes the job: the list-shaped fields get add/remove-row editors, so the raw-YAML hatch is now truly last-resort.

✨ Features

feat(config): profile array editors
- server/lib/routes/content.mjs — PUT /api/profile now also accepts an { arrays: { … } } payload (alongside / combinable with { fields }). Allow-listed paths: target_roles.primary + narrative.superpowers (string lists), target_roles.archetypes (name/level/fit), narrative.proof_points (name/url/hero_metric). Object rows keep ONLY allow-listed sub-keys (injected keys dropped); empty rows dropped; an emptied list deletes the leaf. Same merge-not-replace invariant — scalars, unknown keys, and untouched arrays survive.
- public/js/views/config.js — 4 collapsible add/remove-row editors (string-list rows; object rows with per-sub-key inputs). Save sends { fields, arrays } in one request.
- i18n: 6 new keys × 8 (config.pfPrimaryRoles/Superpowers/Archetypes/ProofPoints/AddRow/RemoveRow).
audit(config): WS6.2 API-keys tab — verified server KNOWN_KEYS (ANTHROPIC_API_KEY/MODEL, GEMINI_API_KEY/MODEL, PORT, HOST) ≡ client FIELDS. Every recognized .env key already has its own labeled input. No gap — no code change.

📝 Documentation

help-bundle §2 × 8 locales — Profile-tab section documents the array editors (add/remove rows, drop-empty, merge-not-replace).

🧪 Tests

test(config): tests/profile-array-editors.test.mjs — 7 cases: string-array merge preserves scalars+unknown-keys, object-array allow-list+drop-empty, empty→leaf-removed, proof_points round-trip, unknown-array-path 400, combined fields+arrays, arrays-only request. 597 → 604.

Verification

$ npm run test:ci
# 604 / 604 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.35.0

[1.34.0] — 2026-05-17

WS5 — one-click Auto-pipeline screen (#/auto).

The v1.15 dashboard auto-pipeline was a transient modal. Promoted to a dedicated, linkable page with senior-UX scaffolding.

✨ Features

feat(auto): dedicated #/auto screen (public/js/views/auto.js) — paste one job URL, one click runs the full chain (validate → fetch → evaluate → save report → append tracker) via the existing POST /api/auto-pipeline SSE contract. Senior-UX:
- Single primary CTA; Enter in the URL field also runs it.
- Live vertical stepper as an ordered list with aria-current="step" on the running row + a polite role="status" live-region announcing every transition (comprehensible without sight).
- On success the result card deep-links to the saved report (#/reports/:slug · N/5) and the tracker — next action one click away.
- Failed step marked red with message; CTA re-enables for fix-and-retry without reload.
- No API key → manual mode: steps collapse to a copy-the-prompt card (no spend).
- Linkable: #/auto?url=<encoded>&go=1 opens + auto-starts.
- Sidebar entry (✨ Auto-pipeline, after Pipeline); dashboard ✨ button now routes here (single coherent flow; the window.AutoPipeline modal helper stays for backward-compat).
- i18n: 14 new keys × 8 locales.

📝 Documentation

help-bundle §1 × 8 locales — new “One-click Auto-pipeline (#/auto) — the 21-step shortcut” subsection (full step list + a11y + manual-mode + deep-link behaviour). H2 count unchanged (17) — added as a ### subsection, no parity-gate churn.
README × 8 — Auto-pipeline headline feature bullet.

🧪 Tests

test(auto): tests/auto-screen.test.mjs — 8 cases: route registration, POST+SSE-drain transport, a11y scaffolding (aria-live + aria-current), manual-mode card, success deep-links, index.html script+nav wiring, dashboard→#/auto routing, 14 i18n keys × 8 locales parity.
589 → 597 unit + acceptance (+8).

Verification

$ npm run test:ci
# 597 / 597 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.34.0

[1.33.0] — 2026-05-17

WS4 — full parent career-ops 1.8.0 feature-parity audit + location_filter.

Audited every user-facing parent commit v1.7.0→v1.8.0 (see .planning/.../PARENT-PARITY.md). One real GAP found and closed; everything else is FLOW (parent script we shell out to / inline via bundleProjectContext), CLI-ONLY (parent Go TUI), or N/A.

✨ Features

feat(scan): portals.yml location_filter parity (parent #570) — parent’s scan.mjs gained an optional location_filter block; web-ui runs its OWN in-process en-scanner/ru-scanner (they do NOT shell out to parent scan.mjs), so it did not flow through. New server/lib/location-filter.mjs mirrors the parent buildLocationFilter semantics verbatim (no key → pass-all; empty location → pass; block precedence over allow; allow empty → pass; allow non-empty → match ≥ 1; case-insensitive substring). Wired into both scanners’ post-fetch filter step (after title/negative, before dedup/persist). Config-driven via portals.yml top-level location_filter: — no UI needed.

📝 Documentation

help-bundle §5 (Portals) × 8 locales — new location_filter subsection with the worked allow/block example + exact semantics + the “top-level key, sibling of title_filter” note.
.planning/.../PARENT-PARITY.md — full classification table of the 20+ parent commits (GAP / FLOW / CLI-ONLY / DOCS / N/A) with rationale per row.

🧪 Tests

test(scan): tests/location-filter.test.mjs — 8 cases: no-filter pass-all, empty-location pass, block precedence, allow-empty pass, allow-match-required, case-insensitivity, malformed-config safe-pass, exact parity with parent’s portals.example.yml worked example. Scanner-integration regression: 51 en/ru/scan tests green (wiring didn’t regress).
581 → 589 unit + acceptance (+8).

Parity outcome

career-ops v1.8.0 feature-parity complete as of web-ui v1.33.0. Deferred (PLAN R4 scope-guard): #341 Turkish as a 9th UI locale — parent added TR mode templates (template-side); a 9th web-ui locale is a dedicated future phase. Optional/LOW backlog: #602 explicit Greenhouse hostname allowlist (already covered by web-ui’s safe-fetch + isValidJobUrl envelope; slug-based URL construction, not raw hostname).

Verification

$ npm run test:ci
# 589 / 589 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.33.0

[1.32.0] — 2026-05-17

#/config Profile tab — raw-YAML blob → field-by-field form (WS1).

Before v1.32.0 the Profile tab was a single monospace textarea where name, email, location, archetypes, compensation and every custom key lived in one undifferentiated YAML blob — the exact pain the user called out (“всё в одной куче”). It is now a structured form.

✨ Features

feat(config): Profile field-form with merge-not-replace save
- public/js/views/config.js — 14 modeled scalar paths grouped into 3 collapsible sections: Candidate (full_name/email/phone/location/linkedin/github/portfolio_url/twitter), Narrative (headline/exit_story), Compensation (target_range/currency/minimum/location_flexibility). Full-name client-side required-check before save.
- server/lib/routes/content.mjs — PUT /api/profile gains a { fields: { "candidate.full_name": … } } payload. Server reads the existing config/profile.yml, sets/clears only the allow-listed leaves, re-serializes the whole object. Arrays the form doesn’t model (target_roles.archetypes, narrative.proof_points, narrative.superpowers) and any custom keys survive the round-trip untouched — the load-bearing invariant (PLAN R1). Empty field → leaf deleted (no phone: "" residue). Allow-list rejects unknown dotted paths; identity gate (full name required) preserved; corrupt existing YAML → 409 (refuses to clobber, routes user to raw editor).
- Raw-YAML escape hatch retained as a collapsed Advanced <details> — the pre-1.32 full-file editor, unchanged ({ yaml } path, replaces whole file, preserves comments). For nested-array edits or comment preservation.
- i18n: 23 new config.pf* / config.profile* keys × 8 locales.

📝 Documentation

help-bundle §2 (App settings) + §3 (Profile) × 8 locales — rewritten for the 3-tab layout + field-form description + the comment-loss tradeoff + Advanced raw-YAML disclosure.

🧪 Tests

test(config): tests/profile-field-form.test.mjs — 7 cases: arrays + unknown-key survival across a field-merge round-trip (the R1 gate), empty-field leaf-deletion, unknown-path rejection, no-full-name rejection, corrupt-YAML 409 guard, legacy raw { yaml } path still works, value trimming.
574 → 581 unit + acceptance (+7).

🔄 Migration

No user action. First Profile-tab save after upgrade merges into the existing file; comments are dropped on a field save (use the Advanced raw editor to keep them). All data keys preserved.

Verification

$ npm run test:ci
# 581 / 581 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.32.0

[1.31.0] — 2026-05-17

Parent career-ops 1.8.0 sync — #/batch exposes --model + --start-from.

The parent project bumped 1.7.1 → 1.8.0. The user-relevant code delta: batch-runner.sh gained --model NAME (#504) and --start-from N. web-ui surfaces both on #/batch, completing flag parity (the runner’s full surface — --parallel --dry-run --retry-failed --start-from --max-retries --min-score --model — is now reachable from the browser).

✨ Features

feat(batch): surface --model + --start-from
- public/js/views/batch.js — new Model text input (optional; placeholder “Claude Max default”) and Start from # numeric input (optional; min 1). Appended to the run-control row next to Max retries.
- server/lib/routes/batch.mjs — GET /api/stream/batch?model=…&startFrom=…. Defense-in-depth (same posture as v1.28.0 --max-retries): model accepted only if it matches ^[A-Za-z0-9.\-]{1,60}$ (rejects shell-meta even though spawn() is arg-array — belt-and-suspenders); startFrom accepted only as an integer 1..100000. Bad values silently dropped, not 400’d — the UI input is the soft contract.
- i18n: batch.modelLbl, batch.modelAria, batch.startFromLbl, batch.startFromAria × 8 locales.

📝 Documentation

help-bundle §14 (Batch evaluate) × 8 locales — --model + --start-from documented in the flag list with the #/batch field-name mapping.
Parent career-ops 1.8.0 analysis recorded. Other parent deltas assessed as no web-ui code change required: interview-prep audience split (#489) flows through bundleProjectContext automatically; dashboard/ is the parent’s Go/Bubble-Tea terminal TUI (browser equivalent is our SPA — not integrated by design); modes/tr/ Turkish templates are parent-side (a 9th web-ui UI locale is a separate deferred decision); parentVersion auto-reports 1.8.0 via health.mjs runtime read.

🧪 Tests

test(batch): tests/batch-model-startfrom.test.mjs — 7 cases: model pass-through, shell-meta charset rejection (injection guard), empty-model drop, startFrom pass-through, out-of-range drop (< 1), non-integer drop, combined-with-other-flags coexistence.
567 → 574 unit + acceptance (+7).

🔄 Migration

No user action. The next 🌐 Batch run with the Model / Start-from fields blank behaves exactly as before (Claude Max default, start at offer #1).

Verification

$ npm run test:ci
# 574 / 574 · ✓ no .also( leftovers · ✓ CHANGELOG parity: all 8 locales at v1.31.0
$ curl -sN 'http://127.0.0.1:4317/api/stream/batch?model=claude-sonnet-4-6&startFrom=5' | grep -m1 '"args"'
# data: {"script":"batch-runner.sh","args":["--model","claude-sonnet-4-6","--start-from","5"]}

[1.30.0] — 2026-05-14

#/scan results paginator — replaces the v1.12.0 “first 200 of N” truncation.

✨ Features

feat(scan): paginate over full filtered result set (public/js/views/scan.js) — pre-v1.30 the scan results table was hard-capped at the first 200 filtered rows with a footnote saying “Showing first 200 of N”. Rows 201..N were unreachable from the UI; the user had to re-tune title_filter.positive in portals.yml to narrow the set if they wanted to inspect later rows. v1.30.0 swaps the cap for UI.paginate (the same helper that drives #/tracker, #/reports, #/activity).
- PAGE_SIZE = 200 preserves the prior visual density per page.
- The FULL filtered set is sorted first (boost-to-top is stable across pages), then page-sliced — so a boosted row that lands on page 2 still appears at the top of page 2, not buried.
- Filter input (text / source / remote / scope / chips) calls pager.reset() so the user lands on page 1 of the new filter result.
- pager.controls(visible, total) renders « ‹ N-M of K › » with disabled-state buttons when on first / last page. When total ≤ pageSize, the controls show only the item count (clean for small datasets).
Stale scan.shownTop i18n key removed from public/js/lib/i18n-dict.js (× 8 locales — no longer referenced).

🧪 Tests

test(scan): tests/scan-paginator.test.mjs — 9 cases across three layers:
- Static-source canaries (7): scan.js declares PAGE_SIZE = 200; wires UI.paginate({ pageSize: PAGE_SIZE, onChange: …renderResults… }); resets pager on filter input; sorts the FULL rows.slice() into sortedAll BEFORE paginating; uses pager.slice(sortedAll); appends pager.controls(sorted.length, rows.length) after the table; no longer contains the pre-v1.30 rows.slice(0, 200) truncation. i18n-dict.js does NOT carry the stale scan.shownTop key. api.js still exports the paginate() helper with .slice / .controls / .reset surface.
- Pure-logic paginator table (1): replicates clamp+slice rules and exercises 6 boundary cases — page 0 of 550 → 200 rows starting at 0; page 1 → 200 rows starting at 200; page 2 → 150 rows starting at 400; overflow page=99 → clamp to last valid page (2); filter-narrow to 5 rows while on page 2 → clamp to page 0 returning the 5 rows; empty set → page=0, empty slice.
- Summary computation (1): mirrors start = page * pageSize + 1; end = min(total, start + visible - 1) from api.js paginate().controls(). Verifies the displayed range across all 3 pages of a 550-row dataset.
558 → 567 unit + acceptance (+9 new).

🔄 Migration

No user action needed beyond updating to v1.30.0. The next scan that produces > 200 filtered rows will surface the paginator below the results table. Smaller scans see only an “N items” hint (unchanged from UI.paginate’s established behaviour in tracker / reports / activity).

Verification

$ npm run test:ci
# 567 / 567
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.30.0

Manual smoke after redeploy: run a scan that produces > 200 filtered rows, navigate #/scan, scroll past the results table, click › to advance to page 2 — should show rows 201..400 instantly (no server round-trip; pure client-side slicing of the already-fetched result set).

[1.29.2] — 2026-05-14

Hot-fix: 🌐 Scan with source=both only ran the EN phase. RU phase was silently dropped.

🚑 Critical hot-fix

User-reported symptom: clicking the unified 🌐 Scan button (which calls runScanAll() → GET /api/stream/scan?source=both) returned only the EN ATS phase. The console showed ▶ ATS scan (Greenhouse + Ashby + Lever) … ✓ ATS done · NEW=0 and then stopped. No RU phase output ever appeared, even with all 5 RU adapters listed in russian_portals.sources.

Root cause — public/js/api.js:156 closed the EventSource on the first done event:

if (ev === 'done' || ev === 'error') es.close();

But server/lib/routes/scan.mjs::driveOne emits done once per phase, and the source=both branch drives two phases sequentially. The client closed after the EN done, the server detected res.on('close') → AbortController.abort() → the RU phase started but was immediately cancelled.

Fix — multi-phase SSE contract:

Server (server/lib/routes/scan.mjs): driveOne now accepts a final param (default true). The done payload carries final: <bool>. The source=both branch passes final: false to the first phase, final: true to the second.
Client (public/js/api.js:148-172): stream(...) closes the EventSource on done only when data.final !== false. Backward-compatible: legacy single-phase producers (/api/stream/batch, /api/stream/pdf*, etc.) don’t set final, so the behaviour is unchanged. Close on error remains unconditional.

🧪 Tests

test(scan): tests/scan-stream-multi-phase.test.mjs — 11 cases covering both server-emitted SSE contract and client decision logic:
- SSE contract (6 cases): source=ats → 1 done (final:true); source=regional → 1 done (final:true); source=both → 2 done events with final:false then final:true; source=both → 2 start events in en-scanner/ru-scanner order; static canary that the pre-v1.29.2 unconditional close pattern is gone; static canary that the v1.29.2 data.final !== false guard is present.
- Functional proof of fix (3 cases): source=both actually emits the RU-phase banner line (proves the body runs, not just empty shells); ru-scanner start arrives AFTER en-scanner done (ordering); dryRun=1 does NOT modify data/pipeline.md (no phase secretly flips writeFiles).
- Pure-logic close-decision table (1 case, parametrized over 11 inputs): mirrors the api.js branch in JS — covers done with final:false / true / undefined / null / 0 / 'false', plus error with payload/null, plus start / log (never close).
- Bug-forensics (1 case): simulates the pre-v1.29.2 client by cancelling the response stream after the first done and verifies the server’s res.on('close') abort handler is still intact (documents the pre-fix mechanism for future readers).
547 → 558 unit + acceptance (+11).

🔄 Migration

No user action needed beyond updating to v1.29.2. The next 🌐 Scan will run both phases.

Verification

$ npm run test:ci
# 553 / 553
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.29.2

# Manual smoke — both phases should now emit:
$ curl -sN "http://127.0.0.1:4317/api/stream/scan?source=both&dryRun=1" \
    | grep -E '^event:|"final":'
# event: start                                        ← en-scanner
# ...
# event: done                                         ← phase 1 of 2
# data: {"code":0,"counts":{...},"errors":0,"final":false}
# event: start                                        ← ru-scanner
# ...
# event: done                                         ← phase 2 of 2 (final)
# data: {"code":0,"counts":{...},"errors":0,"final":true}

[1.29.1] — 2026-05-14

Detailed user-facing config guide for the 5 RU portals in help-bundle §5, all 8 locales.

📝 Documentation

docs(help): §5 "Configuring Russian portals — detailed setup guide" (docs/help/<locale>.md): new ### subsection within Portals & sources covers the end-user config flow:
- 5-row source-inventory table with auth + geo restrictions per adapter.
- Step 1 — locate portals.yml + bootstrap from template.
- Step 2 — full 5-source russian_portals: YAML example.
- Step 3 — tuning queries, area, per_page, only_remote.
- Common pitfalls — negative-list collision (with worked example showing the fix).
- How to disable a single source without losing data.
- How to verify via 🌐 Scan + the per-source SSE log line shape.
Universal YAML/code blocks shared across locales; prose translated for ES / PT-BR / KO / JA / RU / ZH-CN / ZH-TW.
§17 (“How to add a new portal”) was the developer flow shipped in v1.29.0; §5 is the user flow shipped in v1.29.1.

🧪 Tests

test(help): tests/help-ru-config-section.test.mjs — 7 cases asserting every locale’s §5 contains the 5-source YAML, the negative-list collision fix, the disable-one-source example, the 5 adapter labels in the verify block, the 5-row inventory table, the HH_USER_AGENT env-var reference, and the 17-H2 parity contract held after the edit.
540 → 547 unit + acceptance (+7).

Verification

$ npm run test:ci
# 547 / 547
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.29.1

# Manual smoke (after redeploy):
$ for lc in en es pt-BR ko ja ru zh-CN zh-TW; do
    curl -fsS "http://127.0.0.1:4317/api/help/$lc" \
      | python3 -c "import sys,json; print('$lc:', 'YES' if 'trudvsem' in json.load(sys.stdin).get('markdown','') else 'NO')"
  done
# every line ends in "YES" — the §5 expansion mentions trudvsem in every locale.

[1.29.0] — 2026-05-14

Russian-portal scanner expanded from 2 to 5 sources. Source registry + dynamic dropdown. New help-section explaining how to add a 12th.

✨ Features

feat(scan): 3 new RU portal adapters — Trudvsem, GetMatch, GeekJob (server/lib/sources/):
- trudvsem.mjs — Russian government open-data API (opendata.trudvsem.ru/api/v1/vacancies). No auth, no IP gate. Normalizes the documented v1 JSON shape.
- getmatch.mjs — tech-focused RU HTML board. Defensive regex parser, returns [] on parse miss (never throws on healthy 200).
- geekjob.mjs — same pattern as GetMatch. Handles article and div-wrapped card variants.
feat(scan): source registry — single source of truth for every adapter (server/lib/sources/registry.mjs): one array of { value, label, region, configKey } records, consumed by the scanner dispatcher, the GET /api/scan/sources endpoint, and the SPA’s source-filter dropdown. Adding a 12th adapter = one entry here + one adapter file + one row in RU_DISPATCH. The pre-v1.29 three-place drift (hardcoded dropdown / hardcoded if-chain / hardcoded default) is gone.
feat(api): GET /api/scan/sources (server/lib/routes/scan.mjs): returns the canonical source list with Cache-Control: max-age=60. The SPA fetches this on #/scan mount and rebuilds the source-filter dropdown dynamically.
feat(scan-ui): dynamic source-filter dropdown (public/js/views/scan.js): on view mount, fetches /api/scan/sources and paints <option> entries. Build-time hardcoded fallback list survives if the endpoint is unreachable. The filter chip in #/scan now lists 11 sources (6 EN ATS + 5 RU).
feat(ru-scanner): default = 5 sources, dispatcher loop generalized (server/lib/ru-scanner.mjs):
- Default russian_portals.sources (the value used when portals.yml omits the array) now pulls from registry.mjs::RU_CONFIG_KEYS — 5 sources, not 2.
- Pre-v1.29 the dispatcher had two hand-written if (cfg.sources.includes('hh')) / if (cfg.sources.includes('habr')) blocks. v1.29 replaces them with a single loop over RU_DISPATCH that’s keyed by the registry. Adding a sixth source = no scanner-loop edit.

📝 Documentation

docs(help): new §17 "How to add a new job-portal source" × 8 locales (docs/help/<locale>.md) — full English step-by-step (adapter template for API + HTML patterns, registry entry, dispatcher wiring, mocked unit test, portals.yml enablement); 7 locale versions with localized prose + universal code blocks + cross-link to the EN canonical text for the full pitfalls table.
docs(help): §5 + §7 updated for 5 RU sources × 8 locales — russian_portals.sources example now reads ["hh", "habr", "trudvsem", "getmatch", "geekjob"]; the Source-dropdown description names all 5.
Help-bundle section count: 16 → 17 (CI parity contract bumped accordingly).

🧪 Tests

test(sources): tests/sources-trudvsem.test.mjs — 6 cases: normalization, удалённо→remote inference, onlyRemote filter, 5xx propagation, empty results = no throw, null-record safety.
test(sources): tests/sources-getmatch-geekjob.test.mjs — 11 cases across both HTML scrapers: fixture-driven card extraction, nav-anchor skip, empty/null safety, 5xx propagation, onlyRemote filter.
test(scan): tests/scan-sources-endpoint.test.mjs — 4 cases: shape, RU-source list parity (5 entries), EN-source list parity (6 entries), Cache-Control header.
test(ru-scanner): tests/ru-scanner.test.mjs — e2e dispatcher test extended to mock all 5 sources.
test(canonical-docs): 17-H2 parity contract (tests/canonical-docs-coverage.test.mjs) and tests/help-ui.test.mjs — both lifted 16 → 17.
520 → 540 unit + acceptance (+ 20 new). Playwright 32/32 unchanged.

🔄 Migration

For the new RU adapters to fire on your stand, the parent project’s portals.yml must list them:

russian_portals:
  sources: ["hh", "habr", "trudvsem", "getmatch", "geekjob"]
  area: 113
  per_page: 50
  only_remote: false
  queries:
    - "Senior PHP"
    - "Senior Go"

If your portals.yml has NO russian_portals.sources: line at all, the v1.29.0 default kicks in and all 5 sources run automatically. If sources: IS present (as in the pre-v1.29 setup), it’s used verbatim and you must update it manually — the web-ui never edits parent-project files.

Also note: a global title_filter.negative: ["php"] will neutralize every Senior PHP query. The scanner emits a stderr warning at scan time (collision detector from v1.13). Adjust the negative list if you see “0 hits” but expected results.

Verification

$ npm run test:ci
# 540 / 540
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.29.0

$ curl -fsS http://127.0.0.1:4317/api/scan/sources | jq '.sources | length'
11
$ curl -fsS http://127.0.0.1:4317/api/scan/sources | jq '[.sources[] | select(.region=="ru") | .value]'
[ "geekjob", "getmatch", "habr-career", "hh.ru", "trudvsem" ]

[1.28.1] — 2026-05-14

Hot-fix: router 404 on hashes with ?query. HH_USER_AGENT row pruned from health.

🚑 Critical hot-fix

fix(router): strip ?query before route lookup (public/js/router.js) — pre-v1.28.1 Router.go('/evaluate?url=…') produced a hash whose first split('/') segment was the whole "evaluate?url=…" literal, which never matched a registered route → __not_found__ (404). Two reported failures had this single root cause:
- #/pipeline → ▶ button (pipeline.js:145: Router.go('/evaluate?url=' + encodeURIComponent(url))).
- “App settings → Modes” deep link (settings.js:80: href="#/config?tab=modes"). Fix is one line: hash.split('?')[0] before the route-name split. The view itself continues to parse query strings via window.location.hash.split('?')[1] + URLSearchParams (see evaluate.js, config.js).

🧹 Cleanup

fix(health): remove HH_USER_AGENT optional row (server/lib/routes/health.mjs:54) — the row surfaced "unset (hh.ru may 403 from non-RU IPs)" on every Health page render, including for users who never scan hh.ru. The hh.ru adapter falls back to a baked-in UA when the env var is unset; the 403-from-non-RU gate is still documented in docs/help/<locale>.md §16 (troubleshooting) and server/lib/ru-scanner.mjs still emits a stderr hint at scan time. Removing the row reduces dashboard noise without losing any diagnostic.

🧪 Tests

test(router): tests/router-query-string.test.mjs — 3 cases: static-source canary (router.js must split off ? before name lookup), explainer-comment canary (the v1.28.1 fix rationale stays in the source), and pure-logic simulation of current() over four representative hashes (#/evaluate?url=…, #/config?tab=modes, #/reports/abc-123, #/dashboard).
test(health): tests/health-no-hh-user-agent-row.test.mjs — 2 cases: regression guard that /api/health no longer surfaces HH_USER_AGENT row; sanity that adjacent optional rows survived the prune.
515 → 520 unit + acceptance (+ 5 new). Playwright 32/32 unchanged.

Verification

$ npm run test:ci
# 520 / 520
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.28.1

# Manual smoke (after redeploy):
$ open "http://127.0.0.1:4317/#/evaluate?url=https%3A%2F%2Fexample.com%2Fjd"
# (should render the Evaluate view with the URL prefilled — no 404)
$ open "http://127.0.0.1:4317/#/config?tab=modes"
# (should land directly on the Modes tab of App settings — no 404)

[1.28.0] — 2026-05-14

Docs alignment + #/batch --max-retries N UI surface. Closes two open backlog items raised by qa/QA-PROMPT-docs-vs-app.md.

✨ Features

feat(batch): surface --max-retries N control on #/batch (Issue #2) — the canonical batch-evaluate-offers guide documents --max-retries N (default 2) but pre-v1.28 the SPA had no way to set it; users were stuck on the runner default.
- public/js/views/batch.js — new numeric input (1..10), disabled by default, enables only when “Retry failed” is checked. Clears its value when retry is unchecked so an orphaned value can’t leak into the next run.
- server/lib/routes/batch.mjs — GET /api/stream/batch?retry=1&maxRetries=N: parses via parseInt, range-validates 1 ≤ N ≤ 10, silently drops out-of-range/non-integer values (UI is the hard contract; server is defense-in-depth). No-op without --retry-failed.
- i18n: 2 new keys × 8 locales (batch.maxRetriesLbl, batch.maxRetriesAria) in public/js/lib/i18n-dict.js.

📝 Documentation

docs: align AI-assistant list to career-ops.org/docs canonical (Issue #1) — the upstream Quick Start lists Claude Code / Codex / OpenCode / Qwen CLI. Pre-v1.28 we drifted to Claude Code / Codex / Cursor / Gemini CLI / GitHub Copilot CLI, identically across all 8 locales. Resolution: aligned downstream.
- 8 help-bundles (docs/help/<locale>.md) — intro paragraph + comparison-table row both now match upstream canonical. One-liner appended: “other Claude-compatible CLIs work too via the same slash-command surface”, localized per locale.
- 8 READMEs (README*.md) — intro paragraph aligned with the same one-liner. The “Multi-CLI” feature bullet (about web-ui’s own CLAUDE.md/AGENTS.md/GEMINI.md shim files, not about career-ops upstream) deliberately retains its wider list (Cursor / Aider / Gemini CLI) since those CLIs do drive our shims.

🧪 Tests

test(canonical-docs): AI-list regression canaries — 2 new canaries in tests/canonical-docs-coverage.test.mjs:
- Every help-bundle + README must contain “OpenCode” and “Qwen CLI”.
- No help-bundle or README may contain the pre-v1.28 stale phrase “Cursor, Gemini CLI, GitHub Copilot CLI” (Latin or CJK delimiter).
test(batch): tests/batch-max-retries.test.mjs — 7 cases covering: present (maxRetries=3 → flag appended), out-of-range upper (=11 → dropped), out-of-range lower (=0 → dropped), non-integer (=abc → dropped), without retry=1 (always dropped), no-param (runner default 2 wins), combined-with-all-other-flags.
506 → 515 unit + acceptance tests (+ 7 max-retries + 2 AI-list canaries). Playwright 32/32 unchanged.

📒 Deferred (unchanged)

G-005 A-G → A-F report-block realignment — still requires a coordinated commit on the parent santifer/career-ops modes/oferta.md. Tracked in qa/REGRESSION-v1.27.md §11.

Verification

$ npm run test:ci
# 515 / 515
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.28.0

# Manual smoke:
$ curl -sS http://127.0.0.1:4317/api/health | jq '.version'
"1.28.0"

# AI-list canaries — every help-bundle + README mentions OpenCode + Qwen CLI:
$ for f in docs/help/*.md README*.md; do
    grep -L 'OpenCode' "$f" && echo "  FAIL: $f"
  done
# (silent — all 16 files green)

# max-retries flag pass-through:
$ curl -sS http://127.0.0.1:4317/api/stream/batch?retry=1&maxRetries=3 | head -3
# event: error                                      (runner missing in stand)
# data: {"message":"batch/batch-runner.sh not found..."}
# (the path is exercised by tests/batch-max-retries.test.mjs against a stub runner)

[1.27.0] — 2026-05-14

Cosmetic + a11y polish: deduplicate sidebar #/dashboard entry.

fix(sidebar): dedupe #/dashboard entry (public/index.html:25-32) — the brand logo block (<a class="logo" href="#/dashboard">) and the first nav item (<a class="nav-item" href="#/dashboard">) both targeted the same route. Screen readers announced “Dashboard” twice in a row when entering the sidebar, and keyboard users had a useless extra tab stop on a control with no distinct purpose. The brand block now renders as a plain <div class="logo">. The Dashboard nav item remains the sole link.

📒 Deferred (parent-coordinated)

G-005 / PR-B (A-G → A-F report block realignment) — still requires a coordinated commit on the parent santifer/career-ops modes/oferta.md. Tracked in qa/REGRESSION-v1.27.md §Deferred.

🧪 Tests

506 / 506 unit + 32 / 32 Playwright — unchanged. No new test required (cosmetic-only).
CHANGELOG parity gate: ✓ all 8 locales at v1.27.0.

Verification

$ npm run test:ci
# 506 / 506
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.27.0

# Manual smoke (after restart):
$ curl -sS http://127.0.0.1:4317/api/health | jq '.version'
"1.27.0"
$ curl -sS http://127.0.0.1:4317/ | grep -c 'href="#/dashboard"'
1   # was 2 on v1.26.1

[1.26.1] — 2026-05-14

Hot-fix: WCAG 2.5.5 — header .btn height restored to 44 px floor.

🚑 Critical hot-fix

fix(css): restore min-height: 44px + line-height + flex-shrink:0 on .btn (public/css/app.css:391-410) — live Playwright measurement on v1.26.0 found 5 header buttons rendering at 39-41 px (Doctor / Quick scan / Open Pipeline / 🌐 Scan now / ✨ Auto-pipeline a URL) — a WCAG 2.5.5 violation. The fix:
- Adds min-height: 44px to .btn. A stale comment at line 427-430 still claimed this floor was in place, but the declaration itself had been lost between v1.18 and v1.26.
- Adds line-height: 1.2 so the in-block text doesn’t compute the row taller than intended on browsers with looser default leading.
- Adds flex-shrink: 0 + box-sizing: border-box to keep parent flex rows from squashing the button under their own height constraints.
- .btn-sm keeps its existing 32 px floor (small-control exception per WCAG 2.5.5 + 2.5.8 spaced-target) — .btn-sm follows .btn in source order so the override applies.

🧪 Tests

test(wcag): tests/wcag-target-size.test.mjs — 4 static CSS canaries:
- .btn block has min-height: 44px
- .btn block has flex-shrink: 0
- .btn-sm keeps min-height: 32px
- .btn-sm defined AFTER .btn in source order (cascade)
502 → 506 unit (+4) + 32/32 Playwright unchanged.
Live verification via headless Chromium across all 13 sidebar routes — every .btn:not(.btn-sm) measured ≥ 44 × 44 px after the fix.

Verification

$ npm run test:ci
# 506 / 506
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.26.1

# Live Playwright probe (run against server on :4317):
$ for route in /dashboard /scan /pipeline /evaluate /batch /reports /tracker \
                /activity /cv /profile /config /health /help; do
    # use the parent project's playwright to measure
    cd $CAREER_OPS_ROOT && node -e "
      const { chromium } = require('playwright');
      (async () => {
        const browser = await chromium.launch({ headless: true });
        const page = await (await browser.newContext()).newPage();
        await page.goto('http://127.0.0.1:4317/#$route', { waitUntil: 'networkidle' });
        const bad = await page.\$\$eval('.btn:not(.btn-sm)', els =>
          els.filter(b => b.getBoundingClientRect().height < 44).length);
        console.log('$route:', bad === 0 ? 'PASS' : 'FAIL ' + bad);
        await browser.close();
      })();
    "
  done
# Every route → PASS

Out of scope (v1.27.0)

Item	Notes
G-005 — A-G → A-F report block realignment	Still waiting on coordinated parent commit.
G-003 — `README.cn.md` rename	Verified-already-closed: repo has `README.zh-CN.md`, no orphan.
Sidebar duplicate `#/dashboard` (brand logo + nav item)	Trivial cosmetic, zero UX impact.

[1.26.0] — 2026-05-14

Test pyramid (unit → functional → acceptance → e2e) + coverage push to ≥ 93 % line / ≥ 83 % branch. Adopts the 4-tier structure mandated by the v1.25 backlog. Adds 22 new tests targeting the biggest coverage gaps from v1.25’s npm run test:coverage report; introduces the tests/acceptance/ directory for cross-endpoint user-journey tests.

📐 Test pyramid documentation

docs(architecture): TESTING.md describes the 4-tier pyramid (docs/architecture/TESTING.md) — single-source explanation of how the suite is structured and where new tests land:
- Tier 1 (unit) — pure helpers (security, parsers, prompts, file-lock, rate-limit, safe-fetch, env-config). No port binding, no FS beyond stubs, no subprocess.
- Tier 2 (functional) — per-endpoint contracts. createApp() against ephemeral port + temp CAREER_OPS_ROOT.
- Tier 3 (acceptance) — multi-endpoint user journeys. NEW tests/acceptance/ directory.
- Tier 4 (e2e) — Playwright headless Chromium (tests/playwright-{smoke,full-cycle}.mjs, tests/e2e{,-comprehensive}.mjs).
100% line coverage target is explicitly scoped to working functionality — the if (isMain) { … } boot block in server/index.mjs and live-LLM call paths in auto-pipeline.mjs are documented exclusions.

🧪 Tier 2 — Functional gap fills

test(jds): jds-list-create-get.test.mjs — pre-v1.26 only the DELETE path was tested. New file adds 10 tests covering:
- GET /api/jds shape on empty + populated state
- POST /api/jds with explicit slug, auto-generated slug, slug normalization warning
- POST /api/jds empty-body / stripped-to-empty-slug 400 responses
- GET /api/jds/:name body roundtrip, 404 on unknown, traversal rejection
- server/lib/routes/jds.mjs coverage: 61.64 % → 100 % line.
test(auto-pipeline): auto-pipeline-rejection-paths.test.mjs — 10 tests covering every URL rejection branch that doesn’t need a live LLM:
- javascript: / file:// / malformed-string / empty / no-key-body
- SSRF: loopback, RFC1918, link-local IMDS (169.254.169.254)
- mode: 'manual' interaction with rejected URL — error precedes any done event
- server/lib/routes/auto-pipeline.mjs branch coverage: 50.00 % → 59.38 %.

🧪 Tier 3 — Acceptance (NEW)

test(acceptance): jd-evaluate-tracker-flow.test.mjs — first cross-endpoint user journey. Threads 7 endpoints in the order the SPA invokes them:
1. POST /api/jds — save raw JD
2. GET /api/jds — confirm in list
3. GET /api/jds/:name — read body verbatim
4. POST /api/evaluate (manual fallback) — generate prompt
5. POST /api/tracker — add row
6. GET /api/tracker — verify presence
7. GET /api/activity — confirm audit-log entry
Second journey: pipeline-add → preview → tracker → delete cycle.

🧪 Test count

480 → 502 unit + acceptance (+22). 32/32 Playwright unchanged.
npm test now runs tests/*.test.mjs tests/acceptance/*.test.mjs.
npm run test:coverage same. npm run test:ci runs all of the above plus the two CI gates from v1.24.1 (check-no-also-leftovers) and v1.25.0 (check-changelog-parity).

Coverage snapshot

all files                     | 93.66 line | 83.73 branch | 92.91 func
server/lib/security.mjs       | 99.30 line
server/lib/safe-fetch.mjs     | 95.81 line
server/lib/file-lock.mjs      | 93.15 line
server/lib/rate-limit.mjs     | 100.00 line
server/lib/parsers.mjs        | 99.57 line
server/lib/routes/jds.mjs     | 100.00 line  ← was 61.64 in v1.25
server/lib/routes/tracker.mjs | 100.00 line
server/lib/routes/reports.mjs | 100.00 line
server/lib/routes/activity.mjs| 100.00 line

Remaining < 95 % line modules are gated by live-LLM / spawn-mock complexity:

auto-pipeline.mjs (46 %) — uncovered region is the live Anthropic / Gemini evaluate path + report-save + tracker-write SSE flow. Out of scope for the 100 % target per TESTING.md “Goal: 100% line coverage of working functionality”.
batch.mjs (67 %) — uncovered region is streamNodeScript spawn of batch-runner.sh. Out of scope (subprocess mocking gap).
cv-import.mjs (77 %) — uncovered region is pandoc / pdftotext fallback paths when the system tools are missing. Out of scope (env-dependent).

Verification

$ npm run test:ci
# 502 / 502
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.26.0

$ npm run test:coverage 2>&1 | grep '^# all files'
# all files | 93.66 | 83.73 | 92.91

$ ls tests/acceptance/
# jd-evaluate-tracker-flow.test.mjs

Breaking changes

None.

Out of scope (v1.27+)

Item	Notes
Live LLM-path coverage in `auto-pipeline.mjs`	Needs SDK-adapter mock + `withFileLock` stub + report-save stub. Currently 46 % line; would push toward 95 % with the mock harness.
Subprocess-mocked coverage for `batch.mjs::streamNodeScript`	Needs spawn-mock; would push 67 % → 90 %.
`cv-import.mjs` pandoc / pdftotext fallback path	Needs env-injected `which()` stub.
G-005 (A-F report block realignment)	Still waiting on coordinated parent commit.

[1.25.0] — 2026-05-14

Auto-pipeline manual short-circuit + dashboard cosmetic + CHANGELOG parity backfill. Closes G-014 (auto-pipeline ignored mode: 'manual'), G-012 (CHANGELOG parity drift — 6 locales were 2 releases behind), and the dashboard ✨ ✨ double-glyph cosmetic. G-003 (README.cn.md rename) was de-facto already closed — repo only has README.zh-CN.md. G-005 (A-G → A-F report block realignment) requires a coordinated parent-project commit and stays deferred.

🛡️ G-014 — Auto-pipeline `mode: 'manual'` short-circuit

fix(auto-pipeline): G-014 — honour mode:'manual' short-circuit (server/lib/routes/auto-pipeline.mjs:158-195) — pre-v1.25 the route always called an LLM. Passing mode: 'manual' (mirroring /api/evaluate since v1.10.2) was silently ignored, the request hung 1-3 min on Anthropic. Now the handler:
- Accepts mode AND evalMode for back-compat. Either value of 'manual' triggers the short-circuit.
- Emits all 5 SSE stages with status: 'done' / status: 'skipped'. No fetch. No LLM call. No $0.05 per request.
- done payload carries { mode: 'manual', prompt: <buildEvaluationPrompt scaffold>, message } — the SPA can render it like the existing /api/evaluate manual-prompt card.
Closes DoS-risk on HOST=0.0.0.0: previously, even with llmRateLimit capping 10 req/60s/IP, 10 attackers × 10 reqs = $50/min in Anthropic burn. Short-circuit fires before the rate-limit decrement counts toward a real call.
Tests — tests/auto-pipeline-manual-mode.test.mjs: 3 tests confirm (1) mode: 'manual' returns < 2 s with all 5 step keys, (2) even with ANTHROPIC_API_KEY set the short-circuit still fires (the original symptom), (3) legacy evalMode: 'manual' callers keep working.

📝 G-012 — CHANGELOG parity backfill (6 locales × 2 missing releases)

docs(changelog): backfill v1.23.0, v1.24.0, v1.24.1, v1.25.0 in 6 lagging locales — pre-v1.25 only EN had v1.23-v1.24; RU was 1 release behind, the other 6 were 2 releases behind. v1.25 dispatches parallel translation agents (mirrors the v1.23 pattern) to land all four entries in CHANGELOG.{es,pt-BR,ko-KR,ja,zh-CN,zh-TW}.md. RU gets v1.24.0 + v1.24.1 + v1.25.0 (it already had v1.23.0 from the v1.23 cycle).
feat(ci): scripts/check-changelog-parity.mjs gate — fails the build if any locale CHANGELOG’s newest entry is older than the EN canonical. Wired into npm run test:ci. Pre-existing G-012 drift would have caught itself the moment it crossed the EN boundary.

✨ Cosmetic — dashboard double-glyph dedup

fix(dashboard): dedup ✨ glyph in auto-pipeline button label (public/js/lib/i18n-dict.js:219) — dash.autoPipeline carried a leading ✨ in every locale string AND public/js/views/dashboard.js:58 prepended another ✨ in the view. Result: button rendered ✨ ✨ Auto-pipeline …. v1.25 drops the leading glyph from every locale’s DICT entry; the view’s prefix is the single source. Same audit pass swept the rest of the i18n bundle — no other double-glyph patterns found.

🚫 Deferred to a future release

G-005 — Report block A-G → A-F per canonical career-ops.org/docs — requires a coordinated commit in the parent santifer/career-ops project (rewrite modes/oferta.md to emit A=Role, B=CV-match, C=Strategy, D=Comp, E=Personalization, F=STAR — drop C-Risks/G-Legitimacy as separate blocks). v1.25.0 ships the web-ui side ready for the new schema (reports.js already accepts arbitrary block letters since v1.13). Tracked for the next release window when parent + child can land together.
G-003 — README.cn.md → README.zh-CN.md rename — verified during v1.25 prep: repo already has README.zh-CN.md (no orphan README.cn.md anywhere under the worktree). The G-003 finding was stale.

🧪 Tests

477 → 480 unit (+3 from PR-B auto-pipeline-manual-mode.test.mjs).
32/32 Playwright unchanged.
npm run test:ci now runs npm test + check-no-also-leftovers.mjs + check-changelog-parity.mjs.

Verification

$ npm run test:ci
# 480 / 480
# ✓ no .also( leftovers in views/
# ✓ CHANGELOG parity: all 8 locales at v1.25.0

# G-014 — manual mode returns < 2 s even with ANTHROPIC_API_KEY set:
$ ANTHROPIC_API_KEY=sk-ant-test PORT=4317 npm start &
$ sleep 3
$ time curl -sS -X POST -H 'Content-Type: application/json' \
    -d '{"url":"https://job-boards.greenhouse.io/anthropic/jobs/x","mode":"manual"}' \
    http://127.0.0.1:4317/api/auto-pipeline | head -20
# real  0m0.1xx s  (was 1-3 min)
# event: start … event: step (×5) … event: done {"mode":"manual","prompt":"…"}

# G-012 — every locale CHANGELOG carries the v1.25.0 entry:
$ grep -c '^## \[1.25.0\]' CHANGELOG*.md
# 8 files, each → 1

# Cosmetic — dashboard glyph:
$ grep "dash.autoPipeline" public/js/lib/i18n-dict.js
# No leading ✨ in any locale value (view supplies the single glyph)

Breaking changes

None. mode: 'manual' is opt-in; legacy evalMode: 'manual' callers keep working unchanged.

Out of scope (v1.26+)

Item	Notes
G-005 — A-F report block realignment	Needs coordinated parent commit (`santifer/career-ops` rewrites `modes/oferta.md`).
Live execution of QA scenario 31 visual sub-tests	Require browser-driven agent (Claude Cowork). Covered partially by Playwright smoke.
`i18n-dict.js` over 400-LOC target	Translation fixture — exempt by policy. Split would add HTTP requests without a bundler.

[1.24.1] — 2026-05-14

Hot-fix: #/config crash on all 8 locales (G-015).

🚑 Critical hot-fix

fix(config): G-015 — replace removed Element.prototype.also call in config.js (public/js/views/config.js:371) — v1.22.0 N-2 dropped the Element.prototype.also global monkey-patch and migrated cv.js to a free-statement pattern, but missed config.js. The result: #/config crashed at first invocation on every locale with c(...).also is not a function. v1.24.1 applies the same migration pattern from cv.js:188-201 — extract the tree to a const root = c(...), run the activation block on its own, then return root;.

🛡️ CI gate

feat(ci): scripts/check-no-also-leftovers.mjs sweep — walks every file under public/js/views/ and fails the build on any .also( call site (commented references allowed). Wired into the new npm run test:ci script. Future revert of the monkey-patch removal can’t re-introduce the same regression silently.

🧪 Tests

test: tests/config-view-syntax.test.mjs — three guards:
- parse config.js via node:vm.Script (catches syntax-level regressions without needing Playwright)
- assert no .also( survives outside comments
- assert the const root = c(...) / return root; migration anchors are present
474 → 477 unit (+3) + 32/32 Playwright unchanged.

Verification

$ npm run test:ci
# 477 / 477
# ✓ no .also( leftovers in views/

# Browser smoke:
$ open http://127.0.0.1:4317/#/config
# → renders normally, no "is not a function" card. Every locale equivalent.

Out of scope (deferred to v1.25)

G-014, G-012, G-005, G-003 — see v1.25.0 entry below for the bundle.

[1.24.0] — 2026-05-14

Help-bundle content-depth refresh + live execution of QA scenario 31 + RU CHANGELOG end-to-end. Closes both items the v1.23.0 “Out of scope” table deferred to v1.24: the full content-depth refresh of all 8 help bundles from the 5 canonical career-ops.org/docs URLs (was URL-coverage-only since v1.11.x), and the live execution of QA scenario 31 against a running server (was “needs browser agent + LLM credentials” — turned out 6/6 sub-tests are reachable via curl + grep, only the visual sub-tests need a browser).

📖 Help-bundle content-depth refresh

docs(help): refresh en.md from 5 canonical career-ops.org/docs URLs (docs/help/en.md) — pre-v1.24 the EN bundle was 1113 lines and listed the 5 canonical URLs in the front-matter but didn’t expand on them in the body. v1.24 fetches all 5 URLs via WebFetch and deepens the matching H2 sections:
- About career-ops (front-matter) — added principles (data sovereignty, AI-agnostic, human-controlled), “What career-ops is NOT” block, expanded concepts inventory from 6 to 10 rows (added Proof points, JD store, Interview-prep, Batch additions).
- §5 Portals — added canonical bootstrap cp templates/portals.example.yml portals.yml, clarified required vs optional fields per tracked_companies entry.
- §7 Scan — added “no AI tokens consumed” note for Option A, follow-up commands list (apply / contacto / deep / tracker).
- §14 Apply checklist — split into SPA checklist mode vs Manual-vs-Playwright-assisted vs Full CLI flow (canonical 8 numbered steps from /career-ops apply <company> to Submitted. with Evaluated → Applied auto-transition); batch evaluate subsection now has TSV schema table + all 4 flags documented + merge-tracker.mjs --dry-run; Playwright Setup subsection lists install commands, MCP registration, alternative .claude/settings.local.json, headless-by-default note.
16-H2 section parity preserved (CI test help-ui.test.mjs::section-parity asserts exactly 16 H2 sections across all 8 locales).
Each of the 5 canonical URLs appears ≥ 2 times in the bundle (CI test canonical-docs-coverage.test.mjs enforces). Per-URL count after v1.24: what-is-career-ops × 4, scan-job-portals × 5, apply-for-a-job × 3, batch-evaluate-offers × 5, set-up-playwright × 3.
docs(help): translate the v1.24 deepening to 7 non-EN locales — 7 parallel translation agents dispatched. Each target locale (es / pt-BR / ko-KR / ja / ru / zh-CN / zh-TW) gets a refreshed bundle that mirrors the EN structure section-for-section, preserves verbatim code blocks / URLs / file paths / button labels (📁 Upload CV / 🌐 Scan now / ▶ Evaluate / 📄 Generate PDF / 💾 Save) and English abbreviations (CSP, SSRF, TOCTOU, WCAG, ATS, JD, SSE, REST, API), and translates the deepening to publication-grade native technical style in the target language.

🧪 QA scenario 31 — live execution (6/6 PASS)

docs(qa): append last-verified live-execution log to qa/claude-cowork-browser-test-prompt.md — pre-v1.24 scenario 31 was documented but never run against a live server (deferred as “needs browser agent + LLM credentials”). v1.24 ran all 6 sub-tests against http://127.0.0.1:4317:

Sub	Description	Status
31.1	Score thresholds in help bundles	✅ PASS (4.5 × 3, 4.0 × 9, 3.5 × 6 mentions in `docs/help/en.md`)
31.2	Scan workflow endpoints	✅ PASS (`/api/stream/scan-{en,ru}` + `/api/scan-ru/config` → 404; `/api/scan/regional/config` → 200)
31.3	`/api/apply-helper` checklist	✅ PASS (body contains `career-ops apply` + `auto-submit` warning)
31.4	`/api/batch` endpoint	✅ PASS (keys `[exists, runnerExists, raw, rows, additions]`)
31.5	Playwright availability	✅ PASS (`/api/health` reports `Playwright (parent node_modules) ok: true, value: installed`)
31.6	Help-bundle URL coverage (5 URLs × 8 locales)	✅ PASS (40 / 40 ✓)

Visual-only sub-tests (require browser) flagged separately in the QA prompt — they remain runnable via Claude Cowork or npm run test:e2e:browser.

🌐 RU CHANGELOG end-to-end (M-9 follow-up)

docs(translate): CHANGELOG.ru.md retry agent — full body translation (CHANGELOG.ru.md) — the v1.23.0 release shipped with the RU CHANGELOG retry agent still in flight (it had crashed once with a socket error and was re-dispatched). v1.24 picks up the agent’s 1542-line full translation: every entry v1.23.0 → v1.6.0 has a publication-grade Russian body, no more EN-bodied stop-gaps. Style discipline matches the v1.22.0 README quality refresh: “функциональность” / “возможности” / “поведение” replace clunky “функционал”; “через” / “с помощью” replace “при помощи”; active voice over passive; “эндпоинт”, “лимит запросов”, “состояние гонки”, “санитайзинг” as canonical terms; English abbreviations (TOCTOU, CSP, SSRF, WCAG, ATS, JD, SSE, REST, API) preserved.

🧪 Tests

474 / 474 unit + 20 / 20 smoke E2E + 32 / 32 Playwright. Zero behavioral test deltas; every help-bundle CI assertion (16 H2 sections × 8 locales, 5 URLs × ≥ 2 mentions, content floor) still green.

Verification

$ npm test                            # 474 / 474

# Help-bundle deepening:
$ wc -l docs/help/en.md
# ~1270 lines (was 1113 — deepened, not bloated)

$ for url in what-is-career-ops scan-job-portals apply-for-a-job \
             batch-evaluate-offers set-up-playwright; do
    echo -n "$url: "
    grep -c "$url" docs/help/en.md
  done
# what-is-career-ops: 4
# scan-job-portals: 5
# apply-for-a-job: 3
# batch-evaluate-offers: 5
# set-up-playwright: 3

# Scenario 31.6 — 40/40 URL coverage:
$ for lang in en es pt-BR ko ja ru zh-CN zh-TW; do
    echo -n "$lang: "
    for url in what-is-career-ops scan-job-portals apply-for-a-job \
               batch-evaluate-offers set-up-playwright; do
      curl -sS "http://127.0.0.1:4317/api/help/$lang" \
        | python3 -c "import sys,json; print(json.load(sys.stdin).get('markdown',''))" \
        | grep -q "$url" && echo -n "✓ " || echo -n "✗ "
    done
    echo
  done

Breaking changes

None.

Out of scope (v1.25+)

Item	Notes
Live execution of scenario 31 visual sub-tests	Require browser-driven agent (Claude Cowork or `npm run test:e2e:browser`). Out of scope for curl-only execution; covered by existing Playwright smoke.
RU CHANGELOG body translation of older entries (v1.5.x and below)	The retry agent only covered v1.6.0 onwards. Pre-v1.6 entries (`v1.5.x`, etc.) — if they ever existed — remain pre-existing-content.
Visual regression on dashboard screenshots after future SPA changes	`scripts/capture-dashboard-screenshots.mjs` regenerates per-locale PNGs; no automated diff currently.

[1.23.0] — 2026-05-14

i18n split + connection-banner CI fix + localized dashboard screenshots + every backlog stop-gap closed. Ships the three items the v1.22.0 “Out of scope” table flagged for v1.23 (M-9 locale CHANGELOG bodies, N-1 i18n.js LOC split, help-bundle content audit) plus a hot-fix for the smoke E2E test that turned the v1.22.0 main-branch CI red.

fix(client): reset health-poll cadence + visibilitychange eager re-check (public/js/api.js:21-91) — v1.22.0’s M-6 exponential backoff was correct (3 s → 6 s → 12 s → cap 15 s, down from the original cap 60 s) but the in-flight setTimeout was locked to whatever delay was set previously. A server killed at t=0.1 with the first ping at t=3 would fail, double the delay to 6, and the next recovery probe wouldn’t fire until t=9. The smoke E2E’s “Flow 2a: connection banner appears on server down, hides on recovery” waited only 4 s and turned red on main.

v1.23.0 reshapes the polling loop:
- _healthHandle is tracked so setConnectionState(lost=true) can clearTimeout and re-schedule with _HEALTH_MIN. The first recovery probe now fires within 3 s of going down, regardless of what delay was queued.
- _HEALTH_MAX lowered from 60 s to 15 s. Backgrounded tab against a dead server still recovers within one polling cycle when the user comes back; bandwidth savings stay substantial.
- document.addEventListener('visibilitychange') eager-rechecks when the tab regains focus and connectionLost === true — Cmd-Tab back doesn’t wait for the next backoff tick.

🧹 N-1 — i18n.js split (over the 400-LOC target)

refactor(client): split DICT into i18n-dict.js (data) + i18n.js (logic) — pre-v1.23 public/js/lib/i18n.js was 639 LOC. The bulk (lines 23-586) was the DICT translation table — pure structured data. v1.23.0 extracts that to public/js/lib/i18n-dict.js (578 LOC, exempt-from-LOC-rule per CLAUDE.md “Exempt from these limits: generated files, migrations, test fixtures, lock files, vendored code” — translation tables qualify as fixtures), leaving public/js/lib/i18n.js at 86 LOC of pure module logic (well under the 400-LOC target).
Loader contract: i18n-dict.js populates window.__I18N_DICT = { … }, then i18n.js reads it inside the existing IIFE. public/index.html loads them in order — i18n-dict.js before i18n.js — so the IIFE sees a fully-populated DICT at construction time. Missing-dict fallback: every t() call returns its inline fallback or bare key, which surfaces a misconfiguration loudly without crashing the SPA.
Test plumbing updated: tests/i18n-coverage.test.mjs, tests/help-ui.test.mjs, tests/canonical-docs-coverage.test.mjs now run both files through the test VM context (or concatenate their source for the regex sweep), preserving every existing assertion.

🌐 M-9 — Locale CHANGELOG body translations

docs(translate): 7 non-EN CHANGELOG files end-to-end — pre-v1.23 CHANGELOG.{es,pt-BR,ko-KR,ja,ru,zh-CN,zh-TW}.md carried EN-bodied stop-gap notes for every entry from v1.13.0 onwards, with a footer pointing readers at the EN canonical. v1.23.0 dispatches 7 parallel translation agents — one per locale — that rewrite every body to publication-grade technical style in the target language. Stop-gap notes removed. Code blocks, file paths, URLs, commit-message-style strings (fix(security): B-1 — …), env vars, and link labels preserved verbatim across all locales.

🖼️ Localized dashboard screenshots in every README

docs(readme): wire each locale README at its locale-specific PNG — pre-v1.23 only README.pt-BR.md referenced dashboard-pt-BR.png; the other 6 non-EN READMEs still pointed at dashboard-en.png. The screenshots (already captured in v1.22.0 cycle by scripts/capture-dashboard-screenshots.mjs) were present in images/ but unused. v1.23.0 updates every README.{es,ja,ko-KR,ru,zh-CN,zh-TW}.md line 14 to its own dashboard-<locale>.png.

🧪 Tests

Same 474 / 474 unit + 32 / 32 Playwright as v1.22.0. Smoke E2E now 20 / 20 (was 19 / 1 fail on main after v1.22.0 due to the banner-recovery regression; v1.23.0’s reschedule fix closes it).
Three existing tests rewired to handle the i18n split. Zero new test files; zero new assertions deleted.

Verification

$ npm test
# 474 / 474

$ npm run test:e2e
# passed: 20    failed: 0    (was 19/1 on v1.22.0 main)

$ wc -l public/js/lib/i18n.js public/js/lib/i18n-dict.js
#       86 public/js/lib/i18n.js          ← logic, under target
#      578 public/js/lib/i18n-dict.js     ← data fixture, exempt

$ grep -h 'dashboard-' README*.md | sed -E 's/.*(dashboard-[^)]+).*/\1/' | sort -u
# dashboard-en.png    (README.md only)
# dashboard-es.png    dashboard-ja.png
# dashboard-ko-KR.png dashboard-pt-BR.png
# dashboard-ru.png    dashboard-zh-CN.png  dashboard-zh-TW.png

# CHANGELOG translation sanity: each locale file > 200 lines of native content
$ wc -l CHANGELOG.{es,pt-BR,ko-KR,ja,ru,zh-CN,zh-TW}.md | grep -v total

Breaking changes

None. public/index.html now loads two scripts where it loaded one — anyone serving the SPA from a CDN must pick up i18n-dict.js; the script load order is enforced by the order of <script src> tags in index.html. The runtime fallback (empty DICT → t() returns the inline EN fallback) prevents hard crashes when the new file is missing.

Out of scope (v1.24+)

Item	Notes
Help-bundle CONTENT depth refresh from career-ops.org/docs (vs URL coverage)	The 5 canonical URLs already appear in every locale’s help bundle since v1.11.x and Scenario 31.6 in the QA prompt verifies coverage. Content-body depth refresh is a v1.24+ candidate.
Live execution of QA scenario 31 against a running server	Requires browser agent + live LLM credentials. v1.24 candidate.
Per-component touch-target sweep on new mode-page hint paragraphs	v1.22.0 M-1 added `<p class="field-hint">` elements that haven’t been verified against WCAG 2.5.5 min-height in all 8 locales.

[1.22.0] — 2026-05-14

M/L/N backlog clearout + docs alignment + translation quality pass. The entire v1.20.1-BACKLOG.md medium-and-below tier shipped in one release: nine M-items, five L-items, two nits. Plus a docs-alignment audit against the five canonical career-ops.org/docs guides, refreshed system prompts under .claude/ and .github/, and quality-refreshed READMEs in all 7 non-English locales.

🛡️ Security hardening (defense-in-depth)

fix(security): M-4 — entity-aware stripDangerousMarkdown (server/lib/security.mjs) — the pre-v1.22 regex matched <script>, javascript:, on*= as literal substrings. <script>, javascript:, and <img src="data:image/svg+xml,<svg onload=…>"> slipped through. The strip now decodes <, >, &, ", numeric (&#NN;) and hex (&#xHH;) entities before the strip regex runs. Validated by 11 tests in tests/cv-xss-bypasses.test.mjs. Real defense is still the client-side UI.md escape-first pipeline; this hardens the at-rest file.
fix(security): L-2 — bash --noprofile --norc on the batch runner (server/lib/routes/batch.mjs:108) — spawn('bash', [PATHS.batchRunner, ...]) used to inherit the user’s ~/.bashrc. A hostile rc file could influence the run. Now spawn('bash', ['--noprofile', '--norc', PATHS.batchRunner, ...]).

🔒 Resilience

fix(client): M-6 — exponential backoff on health ping (public/js/api.js:22-48) — the disconnected-state poller used to fire 28,800 fetches against a dead server overnight. Now 3 s → 6 s → 12 s → 24 s → 60 s; resets to 3 s on first 2xx recovery. Setup is a setTimeout chain (not setInterval) so each step picks up the new delay.
fix(client): M-5 — Safari private-mode localStorage guard (public/js/lib/i18n.js:572-583) — Safari private-mode throws SecurityError on every localStorage.getItem/setItem. The IIFE-during-load used to fail the entire i18n module, leaving the SPA rendering raw keys. Wrapped both calls in try/catch with the detect() browser-language fallback.
fix(server): M-2 — body-size cap on outbound preview fetches (test + verify) — the v1.21.0 safeGet already streamed chunks and capped at opts.maxBytes. v1.22 adds an explicit regression test in tests/ssrf-redirect-rebind.test.mjs to lock the contract: 100 KB upstream + 4 KB cap → response ≤ 4 KB.
fix(client): L-5 — clear setTimeout on hashchange in scan.js (public/js/views/scan.js:6-22, :113-120) — the post-done 300 ms refreshResults() timer used to leak when the user navigated off #/scan in that window. Handle is now captured and cleared in __cancelActiveScanPoll.
fix(client): L-4 — multi-line SSE data: joiner (public/js/lib/auto-pipeline.js:158-176) — the SSE parser used match() (single-line). Per spec, an event may carry multiple data: lines that the consumer joins with \n. Server currently sends single-line JSON, so the old code worked — but was brittle to any future multi-line payload.

♿ Accessibility

feat(a11y): M-3 — WCAG 1.4.1 redundant cues on score pills + connection banner (public/css/app.css:602-625, :812-822) — score-high / score-mid / score-low used to convey state by hue alone (red/amber/green). Users who can’t perceive hue had no fallback. Each tier now gets a redundant glyph via ::before (✓ / ◐ / ○). Connection banner gets a leading ⚠ glyph in the offline state. Render sites untouched — pure CSS hardening.
feat(a11y): M-1 — inline hint paragraphs for every mode-page field (public/js/views/mode-page.js, public/js/lib/i18n.js) — v1.20.0 wired htmlFor → id for every mode-page field but didn’t carry inline hint copy; only the README walkthroughs documented field intent. v1.22.0 adds 19 hint i18n keys × 8 locales = 152 new translations and the field() builder now renders a <p id="…-hint"> with aria-describedby wiring per field. Screen-reader users hear the hint when the input is focused.
fix(a11y): M-7 — null-guard on UI.el() htmlFor alias (public/js/api.js:194-198) — htmlFor: null used to render literal for="null". One-liner mirror of the fallthrough branch’s v != null && v !== false guard.

🧹 Quality / portability

fix(server): L-1 — parseInt radix in health.mjs + bin/start.sh + bin/setup.sh — parseInt(process.versions.node) without radix triggers a lint warning and is brittle if Node ever ships hex versions. Added 10 everywhere.
fix(server): L-3 — Windows-safe entrypoint check (server/index.mjs:159-163) — import.meta.url === \file://${process.argv[1]}`mishandles drive letters and backslashes on Windows. Replaced withfileURLToPath(import.meta.url) === path.resolve(process.argv[1])`.
refactor(client): N-2 — drop Element.prototype.also monkey-patch (public/js/views/cv.js:188-201) — global DOM prototype pollution. Replaced with a local variable for the tree root.
test(canary): M-8 — 404 regression test for retired /api/scan-ru/config (tests/scan-consolidated.test.mjs) — v1.20.0 retired the alias but added no canary. Three-line addition mirroring the v1.18 retirement tests.

📚 Docs + system prompts

docs(architecture): refresh OVERVIEW + DATA-FLOWS for v1.21+ surface — added safe-fetch.mjs (DNS-pinned GET), file-lock.mjs (per-path mutex), rate-limit.mjs (LLM throttle), and sanitizePathName to OVERVIEW.md. DATA-FLOWS.md gained two new sections: “Outbound URL fetches (DNS-rebind-safe)” and “LLM endpoint rate-limiting”.
docs(readme): security envelope section refresh — README.md “Security notes” now documents every helper in the v1.21+ security envelope (sanitizePathName, safeGet, withFileLock, llmRateLimit, entity-aware stripDangerousMarkdown).
docs(qa): scenario 31 — career-ops.org/docs alignment (qa/claude-cowork-browser-test-prompt.md) — six new sub-tests (31.1–31.6) that verify the UI matches behavior described in the five canonical career-ops.org/docs guides: score thresholds, scan workflow (one button), apply workflow (checklist, not auto-submit), batch workflow (TSV editor), Playwright setup (graceful failure), help-bundle coverage (5 URLs × 8 locales).
docs(translate): README quality refresh × 7 non-EN locales — every non-EN README rewritten to publication-grade technical style in its native language. Common clunky calques replaced; v1.21/v1.22 security envelope mentions added; release/test badges bumped.
docs(system): .claude/PROJECT-CONTEXT.md + .github/copilot-instructions.md — single-file orientation for agents joining a session. Compressed CLAUDE.md, names the v1.21+ helpers, lists common pitfalls.
docs(bin): actualize start.sh / setup.sh / run_all.sh comments — “two deps” → “three deps” (express + js-yaml + multer); “298 tests” → “474+ tests”; parseInt radix added.

🧪 Tests

461 → 474 unit (+13) + 32/32 Playwright unchanged.
New test files: cv-xss-bypasses.test.mjs (M-4, 11 tests).
Extended: ssrf-redirect-rebind.test.mjs (+1 for M-2 body cap), scan-consolidated.test.mjs (+1 for M-8 alias canary).
Zero behavioral test deltas on existing suites — every fix is additive or covered by a new canary.

Verification

npm test                          # 474 / 474
npm run test:e2e:browser          # 32 / 32

# Entity-encoded XSS strip:
node -e "import('./server/lib/security.mjs').then(({stripDangerousMarkdown}) => console.log(stripDangerousMarkdown('&lt;script&gt;alert(1)&lt;/script&gt;')))"
# → '' (no <script> survives)

# Health-ping backoff (open devtools, kill server, watch network panel):
#   3 s → 6 s → 12 s → 24 s → 60 s, then resets on first successful ping

# Score-pill glyph (open #/reports in light + dark theme):
#   .score-high shows ✓ + numeric score
#   .score-mid  shows ◐ + numeric score
#   .score-low  shows ○ + numeric score

# Mode-page hints (#/contacto, etc):
#   <input aria-describedby="mode-contacto-recipient-hint">  ← targets <p id="…">

# Retired alias:
curl -sS -o /dev/null -w '%{http_code}\n' http://127.0.0.1:4317/api/scan-ru/config
# → 404

Breaking changes

None. Every fix is additive or preserves existing endpoint contracts.

Out of scope (v1.23+)

Item	Notes
M-9 — locale CHANGELOG body translations	All `CHANGELOG.{es,pt-BR,ko-KR,ja,ru,zh-CN,zh-TW}.md` v1.13+ entries are EN-bodied stop-gaps. Bulk translation candidate after release cadence slows.
N-1 — `public/js/lib/i18n.js` over the 400-LOC target	Splitting per locale increases HTTP cost without a bundler. Defer until the build-step decision lands.
Help-bundle content refresh from career-ops.org/docs	The five canonical URLs already appear in every locale’s help bundle (since v1.11.x). Scenario 31.6 in the QA prompt verifies coverage. Content depth refresh is a v1.23 candidate.

career-ops-ui v1.267.1 · page 8/10 · mirror of the project changelog for cvstart.org