F0-F2 fully shipped this session; F3 and F5 partially shipped. Each
phase section now states exactly what landed (with commit
references), what was deliberately deferred and why, and what a
future session should pick up next -- so the roadmap stays a source
of truth instead of drifting from the actual repo state.
Notable corrections made while auditing: JobCvMatchService already
reads a structured+text hybrid (the F5 "retarget" line item was based
on a stale assumption, resolved as not-needed rather than deferred);
F1's read-path cutover and F2's pre-existing-row backfill are
explicitly still open, not silently done.
Phase F2 of the Career Workspace roadmap: CvVariant, CvVersion, and
TailoredApplication -- the reference seam. Per the product boundary,
a job application REFERENCES a tailored output; it does not own it.
CvVariant is not job-owned: it survives job deletion (SetNull on its
optional CareerProfile link, not cascaded), can be reused across
applications, and carries its own append-only CvVersion history.
TailoredApplication is the join that links a variant to a job
(cascades with either side, since the link is meaningless without
both).
Rather than shipping empty tables with no consumer, this dual-writes
from both existing TailoredCvDraft save paths (SaveTailoredCvDraft,
UpsertGeneratedTailoredCvDraftAsync via GenerateTailoredCvDraft) --
same pattern as CareerProfile in Phase F1. TailoredCvDraft remains
authoritative for every existing read path; the sync is additive and
never blocks or fails a draft save.
2 new tests: variant/version/link created on first save, same variant
reused (not duplicated) with version incrementing on subsequent
saves. Verified against the real dev DB -- FK dependency ordering
(CareerProfiles -> CvVariants -> CvVersions/TailoredApplications)
holds in both SQLite and MySQL reconciler dialects.
The competitor research flagged ATS transparency as a market gap
worth owning (nobody shows an ATS rating; Canva's canvas layouts fail
72% of parses with no warning). The backend template descriptor
already carried a LayoutFamily/AtsRating pair; the frontend template
picker duplicates that catalog in a hardcoded array and never
displayed it.
Add atsRating to each of the six templates, matching the backend
values (single-column templates: High: sidebar/grid templates:
Medium), and show it as a badge on every template card.
The frontend/backend template catalog duplication itself is a known
gap (frontend never calls GET /profile-cv/templates) -- left as-is
here since rewiring it is Phase F3/F4 scope, not a quick fix.
Extends the interview-prep persistence pattern (previous commit) to
the other two AI-generated per-job outputs that were re-running their
full AI call chain on every tab open: candidate-fit (4 AI calls) and
focus-plan (4 AI calls). Across all three tabs that's 9 AI calls fired
every single time a user revisits a job's AI workspace tabs.
Generalized into AiWorkspaceNote (OwnerUserId, JobApplicationId,
NoteType, ResultJson) rather than duplicating InterviewPrepNote's
per-field-column shape: CandidateFitDto and FocusPlanDto are irregular
and nested (up to 13 fields including a nested guidance object),
where per-field columns would be unreasonable. One table, keyed by
note type, serving both.
Same rules as interview prep: reuse across calls, regenerate when the
attachment selection changes, regenerate on explicit refresh. Frontend
gets the same "Regenerate" button on both tabs.
3 new tests (persist+reuse for both, refresh for candidate-fit).
Verified against the real dev DB.
Interview prep re-ran its AI call every time the tab opened -- flagged
in the product teardown as work evaporating on every re-open (cost,
latency, and non-determinism for no reason). GetInterviewPrep now
persists one note per job application and reuses it on subsequent
reads, only regenerating when the selected attachment context changes
or a refresh is explicitly requested.
- InterviewPrepNote: one row per (owner, job), keyed additionally by
an attachment-selection fingerprint so picking different attachments
correctly triggers a fresh brief without needing an explicit flag.
- GetInterviewPrep gained a `refresh` query param; the frontend adds a
small "Regenerate" button as the explicit escape hatch for when the
underlying job/notes have changed since the note was written.
- Both SQLite (dev) and MySQL/MariaDB (prod) reconciler dialects.
- 3 new tests: reuse across calls, refresh regenerates, attachment
context change regenerates. Verified against the real dev DB.
Four discovery documents backing the Career Workspace redesign:
- cv-builder-competitor-deep-research.md: teardown of Novoresume,
Reactive Resume, FlowCV, Teal, Enhancv, Canva, Resume.io, Kickresume
-- positioning, UX patterns, pricing/trust failures, technical
architecture lessons (esp. Reactive Resume's content/theme
separation and PDF pipeline history).
- cv-builder-product-teardown.md: critical as-is audit of this app's
CV builder -- data model, editor UX, AI workflow, rendering
pipeline, feature gaps -- including the OAuth CV lockout bug fixed
in a prior commit.
- career-workspace-product-strategy.md: product vision, positioning,
personas, core object model, feature roadmap (MVP/V2/future),
AI/monetization strategy, and the first 10 engineering tasks.
- career-workspace-implementation-roadmap.md: the execution plan --
product boundary (Career Workspace is a bounded domain supporting
job tracking, not replacing it), phased sequencing (F0-F6), and the
migration mechanics specific to this repo's raw-SQL schema
reconciler.
The master-CV rewrite preview replaced text without showing what
changed -- the teardown flagged this as the biggest unmanaged AI risk
(a rewrite silently upgrading "assisted with migration" to "led
migration" was invisible). Add a "Show changes" toggle on the rewrite
preview panel that renders a word-level diff (before = current master
text or the targeted section's stored content, after = the AI's
rewrite) instead of the flat replacement text.
Defaults to off: an existing test proved diff-by-default breaks the
familiar plain-text read (word-fragmented spans aren't matchable as
one block), and it's a better UX default regardless -- read normally,
opt into the diff when you want the trust signal.
Uses the `diff` package (word-level diffWords) rather than hand-rolled
LCS; no existing dependency covers this, and it's a solved problem.
CvTemplateRenderer (six hardcoded HTML templates, ~450 lines) had zero
test coverage. Lock in current behavior before any future theme
extraction touches it: renders without throwing and contains the
content it was given, deterministic for identical input, falls back
to ats-minimal for an unknown template id, and HTML-encodes
user-supplied content (regression guard against CV text containing
markup).
Introduces the Career Workspace's bounded data foundation, additive
and backwards-compatible: ApplicationUser.ProfileCvStructureJson
stays the authoritative column every existing read path uses; the new
CareerProfiles/CareerProfileVersions tables mirror it via
ICareerProfileService so future Career Workspace features (variants,
history UI) have real tables to build on rather than starting a
second migration later.
- CareerProfile: one snapshot row per user (Version, ProfileJson).
- CareerProfileVersion: append-only history, one row per save
(upload/rebuild/improve/reprocess/parse), so a profile edit is never
silently lost the way ProfileCvStructureJson overwrites are today.
- Stable item IDs assigned to jobs/education/certifications/projects
on first save and preserved across later saves -- the prerequisite
for CV variants to reference "this job" by identity instead of
array position.
- CvDateNormalizer: best-effort free-string -> "YYYY-MM" parsing for
job/education/certification/project date ranges, kept alongside
(never replacing) the original free-string fields.
- Both SQLite (dev) and MySQL/MariaDB (prod) reconciler dialects,
matching this repo's schema-via-raw-SQL-reconciler convention
rather than EF migrations.
Job tracking is untouched -- this is entirely within the profile/CV
domain per the Career Workspace product boundary.
ProfilePage gated every CV control (upload/rebuild/improve/reprocess/
rewrite) behind isLocal, which is true only for password-authenticated
accounts. Every OAuth signup landed on a CV builder with every button
disabled, even though the backend never restricted these endpoints by
provider (all providers share the same local-scheme session token
after sign-in).
Replace the CV-feature gates with canEditCv (true for any
authenticated user). isLocal is kept for the fields it was actually
meant to protect: password change and provider-managed identity
fields on OAuth accounts.
Dashboard onboarding checklist: a dismissible card with 3 steps (add
CV, import first job, check match score), each linking straight to
where you'd do it. Auto-hides once both CV and a job exist; otherwise
persists per-user via localStorage until dismissed.
Fixes the actual authenticated-landing redirect to /dashboard: my
earlier commit changed App.tsx's inner Shell route for "/", which
turned out to be dead code -- the outer router claims "/" for
LandingPage first, so Shell's own "/" route is never reached on a
direct hit. The real redirect lives in LandingPage.tsx's post-auth-check
navigate() and LoginPage.tsx's post-login nextPath default; both now
point at /dashboard. Verified live: an authenticated visitor hitting
"/" now lands on Dashboard with the onboarding checklist visible,
confirmed via rendered page text and screenshot.
Continuing the product/UX review's deferred items:
- Every top-level view now gets a one-line subtitle under its title
(Dashboard/Jobs/Kanban/Reminders/Correspondence/Gmail review) stating
what that specific view is for, instead of navigation being the only
signal of what each page does.
- Correspondence inbox and Gmail review queue cross-link to each other
instead of being two unexplained flat sidebar items -- kept both nav
entries (renaming/nesting risked breaking muscle memory) but made the
relationship between them explicit in the UI itself.
- Kanban board switches to a horizontal scroll-snap row on phone-width
viewports instead of stacking all 5 columns vertically, which meant
a lot of scrolling to see anything past "Applied".
- Match-score ring gets an aria-label with the actual percentage --
it was two nested decorative CircularProgress elements with no
accessible text. (Keyboard-accessible status changes on kanban cards
were already covered by the existing "..." menu -- no gap there.)
Implements the six "propose first" items from the product/UX review:
- "/" now redirects to /dashboard instead of the empty /jobs table --
a new user's first screen is now an overview with orientation, not
a data table with zero rows and four filter dropdowns.
- Jobs table gets a real first-time empty state (distinct from "no
results match your filters") pointing at Add Job and the bookmarklet,
instead of a bare "No jobs found."
- Match Score card and Candidate Fit tab now each get a one-line
caption explaining what they are and how they differ (deterministic
keyword coverage vs. AI opinion) -- they previously sat side by side
with no explanation of why there are two.
- Google sign-in hint now reflects self-serve signup when
Auth:AllowRegistration is on, instead of always implying you need an
existing linked account.
- Quick Search button now shows its keyboard shortcut (Ctrl+K / ⌘K)
inline instead of being undiscoverable.
Root cause of "Google authentication failed": appsettings.Development.json
had Auth:GoogleClientId set to the literal placeholder
"CHANGE_ME_GOOGLE_CLIENT_ID" while the frontend's .env.development had a
real (already-public, already-committed) client ID -- every Google ID
token's audience check failed against the backend's placeholder. Fixed
by setting the same real client ID on both sides (a client ID is a
public identifier, not a secret, safe to commit -- unlike a client
secret). Also enabled Auth:AllowRegistration in dev so the existing
Google-first self-serve-signup path (auto-create on unmatched verified
email, auto-link on matching verified email -- built during Wave 7) is
actually exercisable locally.
Wired the previously-missing Auth__MicrosoftClientId /
NEXT_PUBLIC_MICROSOFT_CLIENT_ID into docker-compose.yml/.env.example
(distinct from the existing MICROSOFT_CLIENT_ID used for Outlook mail
linking) -- Microsoft sign-in was never deployable, a leftover gap from
when it was built. Fixed a stale env-var name in the Microsoft setup
hint copy (still said REACT_APP_*, predates the Next.js migration).
Removed the per-user accent color picker entirely: it was purely
client-side (localStorage + theme.ts), never touched the backend/DB.
theme.ts now hardcodes a single ACCENT constant; themePrefs.ts drops
get/set/clearAccentColor; App.tsx and SettingsView.tsx drop the
accentColor prop threading. Dead accent-related i18n keys removed from
both locales.
Consolidated Settings' "Account" tab (duplicated GoogleAuthCard, which
already lives on the Profile page) into Profile: moved AuthStatusCard
and EmailProviderConnections there alongside the existing Google/
Microsoft auth cards, so identity/account-linking lives in one place.
Settings drops from 5 tabs to 4 and its General tab uses a consistent
SectionCard layout instead of ad-hoc per-card styling.
Verified: dotnet build/test (177/177) and npm build/test (57/57) both
green; confirmed live against a running dev server that /auth/config
now reports googleEnabled with the corrected client ID, Settings has
no accent controls, and Profile shows the consolidated auth section.
Production deploy has been broken since the Next.js migration merged:
the Dockerfile ran `npm ci` right after COPY package*.json, before the
later `COPY . .` that would bring in .npmrc -- so the legacy-peer-deps
fix for react-scripts' stale TS ^4 peer constraint (added for CI in
dbb1580) never took effect in the actual deploy image, and every
deploy since has failed with the same ERESOLVE error CI hit before
that fix. Copy .npmrc alongside package*.json so npm ci sees it.
Second UI-rework pass. The job workspace mockup's signature element is
a donut "coverage" ring for the deterministic CV match score; the app
had a linear progress bar instead. Replaced with a layered
CircularProgress ring (track + value arc, percentage centered) while
keeping every existing feature (band chip, matched/missing keyword
chips, section coverage) -- this is a pure visual upgrade to the
existing MatchScoreCard, not a feature reduction to match the mockup's
simpler single-panel layout.
Fixed match-score-panel.test.tsx's no-signal-state assertion, which
expected the removed inline "—" placeholder; restored it outside the
ring's conditional render.
First pass of the /frontend-design overhaul against the mockups at
F:\Pictures\website\jobtracker\new. Two highest-leverage gaps from the
backlog note ("dark sidebar, KPI cards, exact status colours"):
- AppShell: nav rail is now a fixed dark navy (#0f172a) regardless of
the app's light/dark theme toggle, matching the mockup's signature
look -- selected item gets an indigo-tinted pill + icon accent,
muted slate text for the rest. Kept icon+label rows (mockup's sidebar
is text-only) since the existing collapsed-sidebar mode depends on
icons; that's a deliberate deviation, not an oversight.
- JobbjaktMark: replaced the briefcase glyph with the gradient
checkmark-in-square mark used throughout the mockups (hero, dashboard,
kanban) -- also fixed a latent SVG gradient id collision across
multiple rendered instances via useId().
- KanbanBoard: mockup uses color sparingly (a small dot in the column
header, a 4px accent on the card's left edge) rather than tinting the
whole column/card background as the previous version did. Reworked
to match; also swapped card title/subtitle order (job title bold,
company/location as subtitle) per the mockup.
Remaining for follow-up passes: Dashboard KPI card layout and the job
workspace (candidate-fit ring, AI summary card) -- both structurally
close already but not yet pixel-matched.
Verified: `next build` clean, all 57 frontend tests green, dark
sidebar confirmed live (computed bg #0f172a) against a running dev
server with light content mode forced.
CI's npm ci (strict peer resolution) rejected the TypeScript 5.9 bump
from the Next.js migration: react-scripts still declares typescript
^3.2.1||^4 as a peer. Local `npm install` didn't catch this -- it
resolves peer conflicts leniently by default; only `npm ci` enforces
them. react-scripts is kept solely as the Jest test runner now (it
doesn't type-check), so relaxing this one peer constraint is safe.
Wave 6. Swaps react-scripts' build/dev tooling for Next.js while
keeping the app's actual routing/rendering model unchanged -- the app
is almost entirely behind auth with no proven SSR/SEO need, so a real
App Router rewrite would touch ~90 files for zero user-visible benefit.
- next.config.js: output:'export' (static HTML+JS, same "single
index.html served by nginx with try_files fallback" deploy as CRA).
- app/layout.tsx + app/page.tsx: root shell ports public/index.html's
<head>, mounts the whole existing App tree client-only (ssr:false)
since it reads window/localStorage during initial render and Next's
static prerender would otherwise execute that on the server.
- Renamed src/pages/ -> src/views/ (Next's Pages Router auto-detects
any `pages/` dir under the app root and tried to build our React
Router page components as its own routes).
- REACT_APP_* -> NEXT_PUBLIC_* across code, .env.development,
Dockerfile, docker-compose.yml build args.
- Replaced the CRA SVGR import (`ReactComponent` from .svg, unsupported
under Turbopack) with a small inline JobbjaktMark component.
- TypeScript 4.9 -> 5.9 (MUI v8's type-checked build needs syntax
4.9's parser rejects; CRA never hit this because babel doesn't
type-check).
- Dropped CRA-only files (index.tsx, reportWebVitals, react-app-env.d.ts,
public/index.html); kept react-scripts as the Jest test runner only
(next/jest migration not needed -- the existing config already works).
Verified: `next build` static export succeeds, `next dev` serves the
landing page and client-side routes (login etc.) correctly, all 57
frontend tests + 172 backend tests still green.
Known caveat: deep-linking straight to a sub-route (e.g. /login) 404s
in `next dev` since there's no server route for it -- the app only
ever mounts at "/". Production is unaffected: nginx's existing
try_files fallback still serves index.html for any path.
Wave 7. Mirrors the existing Google ID-token-exchange pattern (Program.cs
smart-scheme dispatch, JWT bearer scheme, AuthController exchange/link/
unlink endpoints, ApplicationUser fields, reconciler columns) for
Microsoft Entra ID + personal accounts via the multi-tenant "common"
endpoint.
Google/Microsoft sign-in previously only worked for accounts already
linked to an existing local user -- there was no way to actually sign
up via OAuth. Both exchange endpoints now create a new user when no
match is found and Auth:AllowRegistration is true, same gate as
email/password registration.
Frontend: new MicrosoftAuthCard (MSAL popup flow -- Microsoft has no
vanilla-JS equivalent to Google's Identity Services script) wired into
the login page's provider tabs and the profile page's account-linking
section. REACT_APP_MICROSOFT_CLIENT_ID env var, Auth:MicrosoftClientId
config gate on the backend.
Wave 4 hardening. Wrap untrusted CV/job-description/instruction text
in tools/summarizer prompts with explicit delimiters and an
ignore-embedded-instructions rule, since JD text, recruiter emails,
and free-text candidate background all flow into rewrite/normalize
prompts unescaped today.
Match score previously normalized synonyms (JS/Kubernetes/K8s/etc)
only when scanning the job posting, not when checking the CV corpus,
so a CV using an abbreviation the job spelled out never matched.
SkillTagger.MatchesTag reuses the same synonym regex for both sides.
Backlog item 4 (Wave 3, first sub-item). HasResume/HasCoverLetter/HasPortfolio/
HasOtherAttachment were manually-editable checkboxes in EditJobDialog,
completely independent of whether a file was actually attached -- classic
drift: mark 'resume ready' by hand, later delete the resume attachment, flag
stays stuck true forever. User confirmed (asked directly, since removing the
manual-override capability is a product decision, not purely technical):
make them fully computed from Attachments, no manual override.
- AttachmentsController.RecomputeAttachmentFlagsAsync: the single place these
four fields get written now, called after every attachment mutation
(upload, delete, Purpose change) that could affect them. Deliberately kept
as persisted columns (not [NotMapped] computed properties reading the
Attachments navigation collection) -- ~15 query sites build JobApplication
DTOs without .Include(Attachments), so a live-computed property would
silently return false everywhere instead of throwing, the worst kind of
bug. Recomputing at the one write funnel avoids touching any read path.
- Removed HasResume/etc from CreateJobApplicationRequest/
UpdateJobApplicationRequest -- no longer client-settable.
- EditJobDialog: removed the manual checkboxes, kept the (now genuinely
accurate) read-only status chips.
- AddJobModal: stopped sending has*-flags at job-creation time; the
follow-up attachment upload call now sets them correctly via the same
recompute path.
Caught a real bug while testing this: the Purpose-change path recomputed
before saving the Purpose change, so a fresh query missed the pending edit
and the flags never updated. Fixed by committing the mutation before
recomputing.
3 new backend tests (purpose-change sets flag, delete clears flag,
non-primary purpose counts as "other"). 172/172 backend, 25/25 frontend
suites (57 tests) green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backlog item 3 (Wave 2), GmailController slice. Pure mechanical extraction,
no behaviour change:
- GmailDtos.cs: the 26 inline record DTOs, moved to a partial-class file so
every existing GmailController.XyzDto reference (tests included) keeps
working unchanged.
- GmailParsing.cs: the 8 pure static helpers (ApplySyncBoundary,
LooksLikeJobRelatedThread, ToConfidence, ExtractFirstEmail/RecruiterName/
CompanyName/RoleFromSubject, BuildPopupHtml), same partial-class approach.
GmailController.cs: 1200 -> 1022 lines. 169/169 green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backlog item 2. CreateSuggestedJob, RelinkThread, and UnlinkThread each
upserted exactly one GmailReviewDecision by ThreadId but loaded every review
decision for the owner (GmailReviewDecisions.Where(OwnerUserId == x).ToList())
just to linear-scan for the one match. Replaced with FirstOrDefaultAsync
filtered on both OwnerUserId and ThreadId, and added a single-row
UpsertReviewDecision overload alongside the existing dictionary-based one
(still used by the review-queue endpoints, which genuinely need every
decision at once to render the queue).
169/169 green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backlog item 1. The committed ModelSnapshot was empty/stale (21 lines, no
entities) -- `dotnet ef migrations add` scaffolded the whole database from
scratch against it, including the ASP.NET Identity tables, which have never
been created by a real EF migration in this repo (always provisioned via the
raw-SQL reconciler in StartupInitializationExtensions.cs -- see
EnsureIdentityTables' own comment). Applying that diff for real would throw
"table/column already exists" on every environment.
Fix: added migration 20260711181039_SyncModelSnapshot with an intentionally
empty Up()/Down() (see its doc comment) -- it only records itself in
__EFMigrationsHistory and regenerates the snapshot to match the live model,
so `dotnet ef migrations add` produces a real diff for the next schema
change instead of the whole database again. Verified zero side effects
against a copy of the dev DB (only inserts one history row) and against a
fresh empty DB (full migration + reconciler chain runs clean).
That fresh-DB verification surfaced a real, previously-undiscovered bug:
EnsureColumn/EnsureMySqlColumn calls for JobApplications/Correspondences/
Companies/Attachments ad-hoc columns all no-op on a truly fresh database
(the tables don't exist yet -- Migrate() creates them afterward), so a
brand-new deployment's first boot would be missing dozens of columns
(LastReminderEmailSentAt, RecruiterMessageDraft, salary fields, Correspondence
Provider/Subject/Channel/etc.) until the next restart. Also caught: my own b4
change (Correspondence.Provider backfill, already merged) had the same
unguarded-on-fresh-DB bug in isolation.
Fixed by promoting the schema-reconciliation helpers (Exec/HasTable/
HasColumn/EnsureColumn and their MySQL equivalents) from local functions to
class-level statics, extracting the ad-hoc-column blocks into
ReconcileCoreAppColumns/ReconcileCoreAppColumnsMySql, and calling them a
second time right after Migrate() succeeds (reusing the connection already
opened for the CoreSchemaReady check) -- idempotent, so free on every boot
except the first one, where it's now required. No inline logic changed,
pure extraction + one additional call site.
Also added Microsoft.EntityFrameworkCore.Design to JobTrackerApi.csproj
(dotnet-ef tooling requires it on the startup project since EF Core 6+;
previously only referenced by JobTrackerBackend, where the DbContext lives).
169/169 backend tests green. Verified live: full app boot against both a
fresh empty SQLite DB and a copy of the populated dev DB, both clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
b5 of the multi-provider email roadmap (frontend). Adds EmailProviderConnections
-- one card per provider (Gmail, Outlook/Microsoft 365, generic IMAP) showing
connect status and connect/disconnect actions, mounted in SettingsView's
Account tab alongside the existing app-login GoogleAuthCard (a separate
concern: that card is sign-in identity, this is mailbox linking).
Gmail and Microsoft reuse the OAuth-popup + postMessage handshake already
built server-side (mirrors Correspondence.tsx's existing Gmail-connect flow).
IMAP has no OAuth step, so it's a plain host/port/ssl/username/password form
posting to /api/imap/connect, which verifies the connection server-side
before storing it.
Deliberately NOT touched: the Gmail-specific job-candidate-matching/review UI
in Correspondence.tsx and GmailReviewPage.tsx. That backend pipeline
(ListJobCandidateMessagesAsync, GmailReviewDecisions) is still Gmail-only by
design -- generalising it now would mean building fake UI for capabilities
Microsoft/IMAP don't have yet. This is scoped to the piece that's actually
provider-neutral: connect/disconnect status.
Verified live (backend + frontend dev servers): logged in, confirmed all
three /status calls return 200, Gmail connect-url fetch succeeds, IMAP form
submit hits /api/imap/connect and surfaces the expected 400 on a bad host.
Frontend suite: 25 suites / 57 tests green (2 new).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
b4 of the multi-provider email roadmap. The manual/free-text correspondence
entry path already existed (CorrespondenceController.Create) -- this slice
was narrower than the roadmap wording suggests: tag every Correspondence row
with which provider it came from (gmail | manual today; microsoft | imap
once those providers grow an import-into-Correspondence path of their own),
not build a new endpoint.
- Correspondence.Provider (nullable string), reconciled via the existing
EnsureColumn pattern (SQLite + MySQL).
- Idempotent backfill: rows with an ExternalThreadId (historically only
ever written by Gmail import) get 'gmail'; everything else gets 'manual'.
- GmailController.ImportSingleMessageAsync now tags Provider = "gmail".
- CorrespondenceController.Create now tags Provider = "manual".
- Both write sites use a fixed literal, not request input -- no injection
surface introduced. Backfill SQL is static, no interpolation.
148/148 green (147 existing + 1 new CorrespondenceControllerTests; the
GmailController import test gained a Provider assertion in place).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
b3 of the multi-provider email roadmap. Adds ImapConnection model + table
(reconciler pattern, SQLite+MySQL), ImapService (MailKit-backed IMAP client),
ImapProvider implementing the existing IEmailProvider contract unchanged,
and ImapController for credential-based connect (no OAuth — user supplies
host/username/password directly, verified by a live connect before storage).
Scope, documented inline with ponytail: comments:
- INBOX only, no multi-folder support.
- Thread grouping approximates the References/In-Reply-To chain root rather
than the IMAP THREAD extension, which not every server implements.
- External message ids are IMAP UIDs, scoped to the connection's current
UIDVALIDITY.
Security: ran the security-audit skill against this diff (credential
handling + arbitrary-host connect is exactly the class of change the
standing security gate exists for). Found and fixed a real SSRF: the
connect endpoint let an authenticated user point the server at an
arbitrary host:port with no internal-range check, and connect-vs-auth
failure was distinguishable to the caller -- together a working oracle to
fingerprint internal services (loopback/RFC1918/link-local/cloud metadata)
from the server's network position. Fixed with EnsureHostIsExternalAsync
(DNS-resolve + reject internal ranges, re-checked on every reconnect to
close the DNS-rebinding gap) and a single generic failure message that no
longer distinguishes connect vs auth failure. 7 regression tests added.
Dependency: MailKit 4.17.0 (MIT license) on JobTrackerBackend.csproj --
stdlib has no IMAP client; hand-rolling IMAP4rev1 (TLS, SASL, MIME parsing)
would be a large, security-sensitive protocol implementation nobody asked
for, so this is the correct dependency, not a stdlib substitute.
168/168 green (161 existing + 7 new SSRF regression tests; the earlier
14 IMAP feature tests are included in the 161).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
npm run build (Terser minify + fork-ts-checker workers) has now died three
distinct ways on this runner in this session: a printed Terser minify error,
an explicit SIGSEGV, and a fully silent kill with zero output between
'Creating an optimized production build...' and the failure line (OOM/SIGSEGV
signature — process killed before it could flush an error). All three are the
same resource-starved-runner class as the npm ci and dotnet-install flakes
already retried elsewhere in this workflow. Retry once, matching that pattern.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Shell (the single auth guard wrapping every protected route under /*)
redirected unauthenticated visitors straight to /login instead of the home
page, contrary to the intended behaviour. Root cause was one line in
App.tsx's Shell render gate.
Everything else in the guard was already correct: a single centralized
check (no per-page duplication), a loading gate that blocks render until
/auth/config + /auth/me resolve (no flicker-redirect), and 401-triggered
re-checks via the axios interceptor + auth-changed event for expired
sessions mid-session.
Fix:
- Shell now redirects to "/" (home) instead of "/login", still passing
state={{ from: path }} so the originally-requested page isn't lost.
- LandingPage forwards that location.state through to /login on every
"Sign in" CTA (6 call sites collapsed into one goToLogin() helper), so
the home-page bounce doesn't drop the deep-link intent — sign-in still
returns the user to the page they wanted instead of dropping them on
the default /jobs.
- Added LandingPage.authRedirect.test.tsx covering the from-state handoff
end to end (Landing -> click Sign in -> /login receives from). Full
suite: 25 suites, 56 tests green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
b2 of the multi-provider email roadmap. Mirrors the Gmail provider's shape
end-to-end so the two stay structurally interchangeable:
- MicrosoftGraphConnection model + table (reconciler pattern, SQLite+MySQL,
same shape as GmailConnection: encrypted refresh/access token, sync state).
- MicrosoftGraphOAuthService: auth-code + offline-access flow against
login.microsoftonline.com, encrypted token storage via IDataProtector,
message search/thread/detail fetch against Microsoft Graph (conversationId
stands in for Gmail's threadId), attachment listing.
- MicrosoftGraphProvider implements IEmailProvider — no contract changes;
the existing seam was already provider-neutral.
- MicrosoftGraphController: connect-url/oauth/callback/status/disconnect,
mirrors GmailController's OAuth surface exactly (including the popup
postMessage handshake). Job-matching/review endpoints stay Gmail-only for
now, per the roadmap — generalising those needs the frontend provider
picker work, not this slice.
- Registered in DI + IEmailProviderRegistry (multi-registration of
IEmailProvider, resolved by ProviderKey).
- Config: Microsoft:ClientId/ClientSecret/TenantId/RedirectUri, wired through
docker-compose.yml + .env.example alongside the existing Google:Gmail* keys.
- Tests: MicrosoftGraphControllerTests (OAuth lifecycle) +
MicrosoftGraphProviderTests (DTO mapping onto the neutral contract).
147/147 green (135 existing + 12 new).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cesnimda
cesnimda