The assistant opens beside the resume or letter it works on: a third column
at 1280px and wider, in place of the left panel from 1024px, a drawer below
that, and full screen on phones. The bar's button and Mod+J toggle it.
The panel sets up a provider in place, suggests what to ask, streams replies
with Stop and Continue, asks clarifying questions, and shows proposed edits as
change sets with page marks, the outline's "n proposed" pill and the page
caption. Context chips decide whether the next message shares the document and
the posting. Errors keep the message with Retry and Switch model, past
conversations are grouped by document with their outcomes, and the
conversation can be copied as a transcript.
Improve in the rich-text toolbar suggests a stronger verb, a result, a shorter
line or the user's own request for the line holding the caret, and replaces it
only on Replace.
The Agents pages, the builder's assistant sheet and the application copilot
panel are removed; /agent links redirect to the document with the assistant
open. Command palette Ask, Job match's missing terms, Applications' Prepare for
next step and Copy for a job all open the assistant on the right document.
Removes the unused react-resizable-panels and @shadcn/helpers dependencies.
Threads belong to a resume or a cover letter (agent_threads.cover_letter_id)
and count the edits they proposed and the user accepted. The propose_edits
tool rewrites or adds passages by id, resolved against the document as it is
now, and the edit statuses are stored with the message. read_letter joins
read_resume; apply_resume_patch, approvals, revert and archiving are removed.
Message context lets the user leave the document or the linked posting out of
a message; the posting now includes the application's notes. Redis is
optional: without it replies stream directly.
The proposal core (passages, additions, states) moves to
@reactive-resume/resume/proposals so the server and the web app share it.
Adds ai.improve, which suggests a rewrite of one line (stronger verb, a
result, shorter, or the user's own request) and says when it states
something new.
- The page: CSV import/export behind one icon, Add application, a dismissible
nudge for the application waiting longest without a reply (10+ days),
List · Board · Insights · Calendar, search across role, company, contacts
and tags, and Show closed.
- List (the default): grouped by stage in the order that needs you first,
with collapsible groups, the next step (warn when overdue), what was sent,
sorting from the headers and row checkboxes for bulk moves, tags, closing
and deleting. Phones get two-line rows and no board.
- Board: a column per stage; drops and Move to… show the same toast.
- Detail sheet (480 px, full screen on phones): the stage stepper with Move to
next, the next step (the next interview or follow-up) with Edit and Add to
calendar (.ics), what was sent (opening the version sent, read-only, in
History), Tailor a resume and Write a letter, editable salary and source,
contacts, tags, autosaved notes and the activity timeline. Close
application… takes a reason; Delete is in ⋯ and asks first.
- Add dialog: paste a link or posting; its role, company and requirements are
read into editable fields and the posting is saved with the application.
Add, or Add and tailor a resume.
- Insights: how far applications get (from their stage history), how many
heard back and how fast, and tailored against base resumes, above the
existing charts.
- CSV import shows how columns were matched before saving and lets you
download the rows it skips; export adds the closed reason.
- The builder opens History on a version from ?version=.
- Applications end in a `closed` stage with a reason (not selected, withdrew,
accepted another offer, no response). The migration moves `rejected` to
closed + not selected and archived applications to closed, rewrites
`rejected` in their stage history, and ships rollback.sql for older
versions. `archived` stays, deprecated; `rejected` is still accepted as
input and means closed.
- Once an application with a linked resume reaches Applied, the resume is
saved as a "sent" version named after the company, and the application keeps
its id and the resume's Check score then.
- New columns: closed_reason, cover_letter_id (backfilled where exactly one
letter was written for the application), sent_resume_version_id,
sent_check_score and requirements.
- applications.ai.parsePosting reads a pasted link or posting. Links are
fetched on the server: https only, public addresses checked at connect time,
three redirects, 2 MB and 10 s at most. A page's JobPosting data fills the
fields without AI; with a provider, the model reads role, company,
location, salary and requirements.
- MCP application tools take closedReason and coverLetterId.
- ai.atsReview takes optional passages (id, where, text). A suggestion that
rewrites one names it in passageId with the whole new passage, which the
editor offers as a proposal to accept or reject.
- The review prompt fills its placeholders in one pass, so resume text that
looks like a placeholder is sent as it is.
- resume.getById returns applicationId, the job application a resume was made
for, so Check's job match can read its posting.
- Every live check has a category (contact details, dates, layout, section
headings, writing). Reports score the applicable rules, per category too.
- Findings carry a key that uses entry ids instead of array indexes, so it
survives reordering. Keys in the new metadata.check.ignored set findings
aside without counting them against the score.
- TWO_COLUMN_LAYOUT flags a two-column template that prints a sidebar.
- Full-width pages print no sidebar, so sections placed only there are now
reported as never printing instead of passing as main-column content.
- metadata.check also holds job-posting terms hidden as not true; public
viewers don't receive it.
- ats-pdf exports the job-description matcher, spelling variants and the
semantics reader for the editor's job match and parser view.
The dashboard becomes Documents: resumes and saved letters in one
library, in a new app shell (a 240px sidebar, an icon rail on tablets and
bottom tabs on phones) with Documents, Applications, Trash (when it has
items), New (N) and the account row.
- Documents: All / Resumes / Letters with counts, search (/) across
titles, tags and linked applications, sort, grid or list (remembered on
the device), tag chips once tags exist, cards with the real first page,
"Resume · Edited 2h ago", the linked application, lock and "New" badges,
and the first-run screen. A file dropped anywhere on the page imports.
- One menu on cards, rows and right-click/long-press: Open, Rename
(inline), Duplicate, Copy for a job… (resumes) or Link to application…
(letters), Tags…, Lock editing, and Move to Trash with Undo.
- Trash lists days left, with Restore and Delete now (asks once).
- New: import a file in three labelled steps with the result and flagged
fields (resumes, and saved letters' JSON), copy a resume for a job,
start blank (named after its headline until renamed), a new letter, or
a sample resume.
Also:
- /dashboard/resumes and /dashboard/cover-letters redirect to Documents.
- Settings pages sit behind a tab strip until they're regrouped, and the
account menu gains Settings.
- A letter written inside a resume can be copied to Documents from its
entry menu (Q3k), replacing the library's copy action.
- The builder's document menu and the letter editor move documents to
Trash instead of deleting them.
- Dialogs reopened right after closing start fresh.
- Removed: the resume and letter libraries and the old create and import
dialogs.
An imported resume is named from its content (the person's name, else the
headline) instead of a random name. Renaming, tagging or linking a locked
document is refused, as moving it to Trash already was.
Schema: resume gains application_id (the job a copy was made for),
trashed_at and auto_name; cover_letter gains tags, is_locked and
trashed_at.
A new documents router treats resumes and saved letters as one library:
list (live or in Trash, with the linked application), counts, rename,
setTags, setLocked, linkApplication, trash, restore, purge (only from
Trash) and copyForJob, which duplicates a resume, links the copy to the
application and gives the application the copy when it has none.
- resume.delete and coverLetters.delete now move to Trash. Documents stay
there for 30 days and are purged when their owner next lists documents,
so no scheduler is needed.
- Documents in Trash are left out of resume and letter lists, and a resume
in Trash isn't shared: getBySlug and verifyPassword skip it.
- Locked documents can't be moved to Trash; locked letters can't be edited.
- A blank resume created with autoName takes its headline as its name
until someone renames it.
Share & export is one 440px sheet (a full-height bottom sheet on phones)
with Link, Download and History tabs. Share opens Link, the ▾ beside
Download PDF opens Download, the clock opens History, and ⌘⇧S / ⌘⇧E open
their tabs. On desktop the page moves 120px aside so it stays visible.
- Link: the public switch card; the address with a live check (300 ms),
the reason it can't be used and a suggestion, saved only once valid, so
the old address stays live; Copy ("Copied" for 2 s); visitor downloads;
the password (Q3a); Open public page; a QR code; Share via… where the
platform offers it; and views, downloads and time since the last view
over 30 days with a daily chart, or the explanation while the link is
off.
- Download: format cards explained by when to use them, Resume or Cover
letter (with the resume header option) when the resume has a letter,
the file name recruiters see (First-Last-Resume by default, unsafe
characters stripped), a non-blocking note about open Check issues, and
progress inside the button; a failure offers Try again and PDF.
- History: name the current state; a timeline of Now, sessions, named
versions, restores and where the document came from. Picking a version
shows it on the page, read-only and outlined, with its banner; Restore
saves "Before restore" first. Named versions can be renamed or deleted.
Also:
- Autosaves send this visit's session id, and naming or restoring a
version saves pending edits first.
- The one-click PDF and every export use the First-Last-Resume name.
- Resume dialogs stop asking for a slug (it's edited in Share), which also
stops Rename from overwriting a custom slug with one made from the name.
- The public route redirects a renamed resume's old address.
- Removed: the download dialog, the version-history menu and the sharing,
statistics and export sections.
Versions:
- resume_version gains kind (created, import, auto, named, before-restore,
restored, ai, sent), name and session_id. The migration backfills kind
from the old English labels; label stays for API clients.
- Each editor visit sends a session id with resume.update, and its saves
keep one autosave version, refreshed at most every two minutes. Calls
without a session keep the old throttled autosave.
- Creating a resume writes a "created" version and importing an "import"
one, so history is never empty.
- New procedures: getVersion (for previews), createVersion (name the
current state), renameVersion and deleteVersion (named versions only).
- Retention replaces the 30-row cap: autosaves, AI edits and restores
expire after 90 days, pruned when a resume gets a new version (there is
no scheduler, and the Vercel entry skips startup hooks); a cap of 500
autosaves per resume bounds storage.
Slugs:
- resume.checkSlug validates ^[a-z0-9]+(-[a-z0-9]+)*$, reports which of the
user's resumes uses a taken slug, and suggests a free one.
- A changed slug must match the pattern (existing ones keep working until
changed). The old slug is kept in resume_slug_redirect for 30 days, and
getBySlug and verifyPassword resolve it; the response carries the current
slug so clients can redirect.
- create and duplicate make the slug optional and generate a unique one
from the name; duplicate no longer falls back to the original's slug,
which always collided.
The migration also applies the pending drop of the redundant
resume_user_id_index removed from the schema in b953435f2.
Columns, sidebar side, header placement and ATS safety for every template
now live in templateLayouts. The template gallery metadata and the layout
editor read it, and a DOCX test checks the two-column configurations
against it.
Dated entries and roles carry dates (start, end, present, and raw when the
text couldn't be read exactly). The parser moves to the schema package and
reports how each date was written; parseResumeData fills dates, infers the
date format from how dates were typed and rewrites the legacy period/date
text from them, so older clients and API readers keep working. API writes
sync against the stored data, so an edit to the text alone is read back
into dates.
getById and getBySlug return upgraded data, ATS date rules and sorting read
dates, JSON Resume and LinkedIn imports map their dates directly, entry
titles may be empty (drafts aren't printed), and the MCP schema resource is
generated live in place of the stale schema.json.
Icons now draw their ligature from data-icon in a pseudo-element, so icon
names stay out of copied text, find-in-page and text queries. useBreakpoint
reports the design system's mobile, tablet, desktop and wide ranges.
Replace the achromatic shadcn palette with the Desk & Paper tokens (warm
neutrals, one moss accent, danger, warn and info), exposed to Tailwind
under their spec names; the previous names resolve to the new tokens
until every screen is rebuilt. Hard-coded palette classes become semantic
tokens and stage colors follow the spec.
Fonts move to Newsreader, Hanken Grotesk and JetBrains Mono. The theme
gains a System option that follows the operating system live, applied
before first paint, and Base UI now receives the locale's direction.
Primitives follow the component spec: button variants primary,
secondary, ghost, danger and link with a loading state, accent focus
rings on inputs, SwitchRow, semantic badges, segmented and underline
tabs, restyled menus, dialogs, sheets, tooltips and command bar, and a
single bottom-center toast with an Undo action. New: IconButton,
SegmentedControl, RadioGroup and NativeSelect.
Icon renders Material Symbols Rounded at weight 300 from a subset font
that holds only the glyphs listed in packages/ui/src/icons/names.ts.
pnpm icons:build checks every name against Google's codepoints and
regenerates the font; a test keeps the manifest and the list in sync.
Icons are aria-hidden and untranslated, and directional ones mirror in
right-to-left layouts.
Templates tag the views that own the header, each section and each item
with data-resume-node. ResumeDocument's new onPageMap callback reads React
PDF's layout tree after each render and returns page-relative boxes, so the
editor can link lines on the page to entries in the panel. The attribute
stays on layout nodes and never reaches the PDF.
Crowdin shipped an empty Central Kurdish (ckb-IR) catalog with no translated
strings. The locale is not registered in the Lingui config or locale schema,
so remove the file and its PDF section title entry.
* feat(applications): schedule interviews and view them on a calendar
Interviews (screening, technical, behavioral, onsite, other) are stored as
"interview" entries on an application's activity timeline, so an application
can have any number of them and they show in its timeline without a
migration. Each interview has a start date-time (timezone-aware), duration,
and optional location and notes.
- schema: interview timeline entry type, interviewDetailsSchema, INTERVIEW_KINDS
- api: addInterview / updateInterview procedures (delete via timeline entry);
generic timeline updates now only edit text on note entries
- web: Calendar view on the Applications page (month grid, type legend,
upcoming list grouped by day, schedule button with application picker,
per-day "+" and "+N more" popover), Interviews section and interview
dialog in the application detail panel, interview rows in the timeline
and CSV export
- mcp: add_application_interview / update_application_interview tools
- i18n: extract new strings into all locale catalogs (English fallback)
- docs: MCP tool table, scheduling guide section, resume-builder skill
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(applications): keep interview dates in local time and guard generic timeline edits
- Format the timeline date chip for interview entries in the viewer's local
timezone so it matches the interview's date-time label; stage and note
entries still render in UTC.
- Reject interview entries in the generic updateTimelineEntry path. A
day-granular date edit kept the UTC time of day and could move an interview
to the wrong local day; callers are pointed to updateInterview
(update_application_interview). The MCP tool description now says so, and
a service test covers the rejection.
- Associate each interview dialog label with its control via useId/htmlFor.
- Drop the duplicate onInput handler on the date-time input; onChange covers
controlled inputs.
- Give the calendar's per-day schedule button an accessible name that
includes the date, and update the extracted locale catalogs for the new
message.
* [autofix.ci] apply automated fixes
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Amruth Pillai <im.amruth@gmail.com>
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
* feat(import): add LinkedIn data export as a resume import source
LinkedIn's "Get a copy of your data" export ships a ZIP of per-topic
CSVs (Profile, Positions, Education, Skills, Languages,
Certifications). Reading these directly gives structured data without
needing a connected AI provider, unlike the existing PDF/DOCX import
path.
Also fixes a latent bug found while building this: parseJSONResume
(and the new LinkedIn parser) built their result via a shallow spread
of the shared `defaultResumeData` singleton, so assigning into
`result.sections.x` mutated that singleton in place and leaked section
data into the next unrelated import call in the same process. Both now
start from a structuredClone.
* fix(import): escape LinkedIn text, harden zip parsing and date handling
Move escapeHtml and toHtml from the plain-text importer into html.ts and
use them for LinkedIn summary, position descriptions and education notes,
so CSV text is HTML-escaped and line breaks become paragraphs or bullet
lists instead of collapsing into one run-on paragraph.
Only an empty end date now marks an entry as ongoing. Date cells that are
not "Mon YYYY" are kept verbatim, so a finished role with an unexpected
date format no longer reads as "Present".
Unzip only the six CSVs the importer reads, matched by exact file name,
and reject any of them larger than 5 MB. This avoids inflating the rest
of a complete LinkedIn export in the browser and stops Learning_Profile.csv
being read as Profile.csv.
Map LinkedIn's five language proficiency options onto levels 5 to 1,
falling back to parseLevel for anything else. Drop the literal BOM strip,
which TextDecoder already handles.
The import dialog no longer mentions an AI provider in the loading toast
for LinkedIn imports, which are parsed entirely in the browser.
Add a regression test for the JSON Resume importer leaking section data
through the shared defaultResumeData object, plus LinkedIn tests for HTML
escaping, unrecognised end dates, BOM headers, exact file name matching,
language levels and oversized entries.
---------
Co-authored-by: Amruth Pillai <im.amruth@gmail.com>
Bump workspace dependencies to their latest versions and dedupe the lockfile.
The upgrade left stale duplicates in pnpm-lock.yaml that broke the build and tests:
- @deepseek-ai/schemastery resolved to both 3.18.2 and 3.18.4. Both copies declare
the global Schemastery namespace, so dsh-plugin's declaration emit failed with
TS2883 on `Config`. `pnpm dedupe` collapses it to 3.18.4.
- vite's optional tsx peer resolved to 4.23.13 for importers without a direct tsx
dependency and 4.23.15 elsewhere, producing two vitest 5.0.2 instances. Loading
both in one run broke `expect(...).rejects`. Re-resolving tsx unifies the graph.
Audit every animation in the app and shared UI primitives against a
frequency-first motion bar: keyboard and high-frequency actions no longer
animate, remaining motion uses interruptible CSS transitions with shared
easing tokens, and redundant animation code is removed.
UI primitives (@reactive-resume/ui)
- Dialog, alert dialog, popover and tooltip move from tw-animate keyframes
to Base UI data-starting/ending-style transitions; menus, popovers and
tooltips skip motion when opened from the keyboard (data-instant).
- Dialog gains an `instant` prop; the command palette uses it.
- Accordion animates its real panel height; caret rotates instead of
swapping icons.
- Menu backdrop blur moves onto the popup so it no longer snaps in after
the fade; context menus and comboboxes fade only.
- Sidebar collapse uses the strong ease-out curve and snaps on Cmd+B.
- Toast, sheet, checkbox, tabs, toggle, inputs and message scroller get
tokenised easing, correct transition properties and press feedback.
- Tabs no longer squeeze a trigger narrower than its label.
- Spinners keep spinning under prefers-reduced-motion.
Web app
- Remove the default route view transition and page-entrance slides.
- Add EASE_OUT_STRONG for Motion; replace built-in "easeOut" everywhere.
- Switch LazyMotion to domMax so layout and Reorder animations run.
- Builder: consolidate 12 section list files into one ItemsSection,
opacity-only list rows with popLayout, instant Cmd+0, faster dock zoom,
crossfade that no longer dips, transform-based progress bars.
- Dashboard: keep previous results while sorting/filtering, no empty-state
flash, uncontrolled sidebar (no network round trip on collapse), calmer
resume card tilt, no stacked hover wrappers.
- Settings: drop entrance/stagger wrappers and ActionButton.
- Agent: CSS marquee paused on hover; thread switches keep the layout.
- Homepage: fix invalid transition declarations, CSS spotlight drift,
scroll-hiding header without a JS spring, tokenised curves.
- Theme switches change every color at once.
- Remove SSR-only useIsClient guards from the SPA.
Docs: rewrite the DESIGN.md animation section around the new tokens.
* feat(deploy): support Vercel Hobby alongside Docker
* fix(deploy): include PDFKit runtime font assets
* docs(deploy): document Vercel and Docker setup
* docs(deploy): record storage persistence checks
* refactor(deploy): drop scheduled staging cleanup
Staging uploads are deleted after finalization and expired ones are swept
on each new upload, so the Vercel cron job, its route, and CRON_SECRET are
no longer needed. The Deploy with Vercel wizard now asks for two secrets.
* docs(deploy): restructure Vercel guides
Split the Vercel page into a how-to with its environment reference, move the
large RPC staging protocol to an API reference page, and move CI deployment
checks to the contributing section. Point Deploy with Vercel buttons at main.
* chore: remove agent planning records and fix web app description
Delete superpowers plans/specs, ADRs, issue plans, execution briefs, domain
context maps, and Europass research. Describe apps/web as a TanStack Router
SPA served by apps/server.
* refactor(deploy): simplify Vercel support code
- Share one Redis client and key namespace through @reactive-resume/db/redis
for API and auth instead of a second auth-only client.
- Drop the auth seeding retry; the provider already treats concurrent inserts
as no-ops and deployment preparation seeds before runtime.
- Detect staging support from POST /api/storage/stage (404 on Docker) instead
of a separate GET probe.
- Read staged bodies directly; the signed upload already caps their size.
- Close per-subscription Redis connections with disconnect() alone.
- Check Blob health with one list call instead of write/read/delete.
- Remove redundant tsdown onlyBundle list, dead namespace fallbacks, and the
conditional spread in the health status.
* fix(deploy): heal stopped runs with dead owners and keep auth up without Redis
- Run owners refresh a Redis heartbeat until they release their claim. Stop
requests reap the run immediately when the owner has stopped heartbeating,
instead of leaving the thread blocked until the 15-minute TTL reaper.
- Auth and oRPC rate limiters fall back to per-instance memory limits when
Redis errors, instead of rejecting every login or failing requests.
* ci: allow esbuild build for Vercel CLI and register deployment deps with knip
pnpm 12 fails dlx installs with ignored build scripts, so allow esbuild
explicitly. The server bundle keeps @vercel/blob, ioredis, and jose external,
and api/index.mjs is the Vercel Function entry.
* fix(web): send buffered RPC bodies instead of teed streams
Reading a request clone turned the original body into a stream, which
browsers send without inspectable request data and which needs duplex
mode. Send the already buffered Blob for direct requests.
* fix(web): send direct RPC bodies as bytes
Blob request bodies are sent as data pipes, so browser tooling cannot
inspect them. Buffer the original request as an ArrayBuffer and send those
bytes; this restores the e2e save assertions that match on request data.
1.0.9 rejects the Path2D objects pdfjs-dist 6.3.289 passes to fill/clip,
failing every raster test with "Value is none of these types `String`, `Path`".
Adding packages/pdf to CI turned it red on ubuntu-latest for two reasons,
neither of which is a real regression.
Rasterized pixels depend on the host font rasterizer. The chikorita, ditto
and all-template baselines differ on Linux in rasterSha256 alone: every page
count, item count and text coordinate is byte-identical to the macOS-authored
baseline. Compare the portable geometry on every host and the pixels only on
the platform the PNGs were generated on, so the characterization keeps
protecting layout without asserting another machine's antialiasing.
The picture-fit override case rasterizes twice and timed out at Vitest's 5s
default on a CI runner, with the date suite landing at 3.5-4.7s. Give the
package a 30s timeout rather than leaving every rendering test a runner
slowdown away from failing.
Verified on linux/amd64 in Docker: both files pass, 18/18.
The Lapras section marginTop added in c0c658c0 shifted every date marker in
that template down the page, but the date-layout characterization baseline
was not regenerated, so packages/pdf has been failing on main since. The
delta is geometry-only: same page count (2), same text item count (87), same
markers, x and width unchanged; 17 markers move on y and the raster hashes
follow.
Nothing caught it because the workflow ran test:ci for only server and
tooling, leaving 17 packages uncovered. Replace that filter list with the
full workspace run so a package cannot silently lose coverage again, and
move it after the migration and storage steps that the api suite needs.
Serial execution is deliberate. Running the packages in parallel oversubscribes
the runner and starves the PDF rasterization and API rate-limit suites past
their timeouts; 19/19 pass consistently at --concurrency=1.
Patch and minor bumps across the AI provider SDKs (@ai-sdk/*, ai),
@aws-sdk/client-s3, react-email/@react-email/ui, knip and jszip, with
pnpm-lock.yaml regenerated to match.
Also records the audit overrides and patched dependencies in the root
package.json alongside the existing pnpm-workspace.yaml entries.
Amruth Pillai
Syed Ali Abbas ZaidiandAmruth Pillai
PerryLinkandAmruth Pillai
Santhi PrakashandAmruth Pillai
Emanuele TonelloandAmruth Pillai