boardwatch

A self-hosted job radar that reads the fine print. Point it at the companies you care about. boardwatch watches their official ATS job boards, catches new postings early, and ranks them against your profile with an explainable score. It also reads each posting for the hard eligibility requirements that quietly rule people out (visa sponsorship, security clearance, a required degree, years of experience, a location) and flags the ones you could not actually apply to, each backed by the exact sentence it read as evidence. For the leads worth your time it renders a one-page résumé PDF tailored from what you wrote, never adding a claim you did not make, and lays them out in a local web page to work through. Nothing is guessed, nothing phones home, and it all runs on your own machine.

Status: pre-release, under active development. boardwatch does not submit applications: it finds and evaluates postings, and you decide what to do with them. No telemetry. The default install needs no accounts and no API keys (the LLM tiers are opt-in). Your data stays on your machine: a local SQLite file and the folders run writes.

$ boardwatch top 5
 #   Title                            Company     Score   Eligibility   Why
 12  Senior Backend Engineer          Stripe      0.86    no flags      covers 9/11 skills · title · 1d
 7   Software Engineer, Platform      Linear      0.81    check         covers 7/10 skills · title · 3d
 33  Backend Engineer (Payments)      Ramp        0.74    no flags      covers 6/9 skills · 2d
 5   Full-Stack Engineer              Supabase    0.68    no flags      covers 5/8 skills · title · 6d
 18  Infrastructure Engineer          OpenAI      0.61    check         covers 4/9 skills · 4d
2 hidden as ineligible. "no flags" means no catalogued disqualifier was detected, not that you qualify.
41 hidden as non-software roles — see them with --include-non-swe, each with the title text that vetoed it.
→ `boardwatch show <#>` for the full posting and eligibility evidence

(Illustrative output. # is the posting id; pass it to boardwatch show <id> for the full posting, a per-component score breakdown, and the eligibility audit with quotes. "no flags" means no catalogued disqualifier was found, not that you are cleared to apply; "check" means the posting was ambiguous.)

Why boardwatch?

Job boards optimize for their advertisers, not for you. LinkedIn/Indeed bury fresh roles under sponsored noise and stale reposts; paid trackers put a subscription (and their servers, and your search history) between you and postings that are already public.

boardwatch takes the direct route. Greenhouse, Lever, Ashby, Workable, SmartRecruiters, Workday, Eightfold, Oracle HCM, Phenom and Jibe each serve every board they host from a public, keyless endpoint, the same data the company's own careers page renders; Amazon's and Apple's own career sites are read the same way. boardwatch polls those endpoints politely, on your schedule, and tells you what's new since last time.

Honest limits. boardwatch's boards cover companies hosted on Greenhouse, Lever, Ashby, Workable, SmartRecruiters, Workday, Eightfold, Oracle HCM, Phenom or Jibe, plus Amazon and Apple (a large slice of tech, but not everyone, no Taleo/etc. yet). Opt-in discovery lanes (lanes_enabled, off by default) can add postings from listings beyond the ATS boards — LinkedIn guest search, Indeed, hiring.cafe, GitHub new-grad lists and schema.org JSON-LD pages — and are held to a different standard, set out under Responsible use. It reads exactly what those sources expose. Its bundled role knowledge covers software roles: another field can declare its own role families at init, but the eligibility catalog is still software-first. It is pre-release: expect rough edges, and read Responsible use before pointing it at boards you don't own.

Quickstart (a first shortlist in a few minutes)

Prerequisites. Python 3.11–3.13 and pipx (or uv). Scanning and ranking need nothing else. Building résumé PDFs additionally needs two binaries on PATH — tectonic and poppler's pdfinfo:

brew install tectonic poppler                       # macOS
sudo apt-get install poppler-utils                  # Debian/Ubuntu (tectonic: see its docs)

boardwatch doctor reports whether both are present. The Docker image bundles them.

Supported on macOS and Linux. Windows is best-effort — see platform support.

pipx install boardwatch      # isolated, on your PATH (or: uv tool install boardwatch)
boardwatch init              # pick companies, then describe what you are looking for
boardwatch scan              # poll the watched boards (polite, conditional GETs)
boardwatch top               # ranked shortlist
boardwatch show <#>          # one posting: score breakdown and eligibility evidence

boardwatch init is interactive. Pick [1] Starter set to watch 15 well-known boards in one keystroke, [2] Search registry to pick from the bundled catalog, or [3] Paste any provider:slug or board URL. Then paste your résumé text or a short profile, a few target and excluded titles, and locations. The eligibility questions are optional (you can answer them later). Last, name your field for the role filter: software uses the bundled role knowledge, any other name asks for your own role families, and blank leaves the filter off. On the starter set the first scan takes well under a minute (about 30 s and 3,200 postings when this was written), and the first top about two minutes, because it reads and evaluates every new posting once; after that top answers in a second or two.

Two settings most people want next, both in boardwatch profile edit: a target seniority band (entry, mid or senior; the default any leaves the seniority filter off) and target countries (ISO-3166 alpha-3 codes such as USA), without which the country-based location filters stay off.

top remembers what it showed you: a lead it surfaced stays off run's shortlist for seen_ttl_days (7 by default), so the same job is not served twice. top --no-record looks without marking anything seen.

New here? boardwatch guide prints the full command reference, generated from your installed copy. It is written for a coding agent working on your behalf, so it also says what each command reads and writes; boardwatch skill prints a short skill file for that agent to save.

Other install methods

Docker

docker run --rm -it -v boardwatch-data:/data \
  ghcr.io/mit112/boardwatch:latest --data-dir /data init
docker run --rm -v boardwatch-data:/data \
  ghcr.io/mit112/boardwatch:latest --data-dir /data scan
docker run --rm -v boardwatch-data:/data \
  ghcr.io/mit112/boardwatch:latest --data-dir /data top

init asks questions, so it needs -it. Everything lives in the boardwatch-data volume: the database at /data, and your settings, résumé files and role taxonomy under /data/config. The review app (boardwatch web) binds loopback only, so it cannot be reached from outside the container; for the daily pipeline, install natively.

From source

git clone https://github.com/mit112/boardwatch && cd boardwatch
uv sync                      # https://docs.astral.sh/uv/
uv run boardwatch init

The daily pipeline

scan and top answer "what is new and worth a look". boardwatch run does the whole job in one command: it scans, evaluates eligibility, ranks, re-checks that each shortlisted posting is still live, and renders a tailored one-page résumé PDF for every lead it puts in the apply lane. boardwatch web then opens those leads in a local review page.

It needs a résumé to tailor, which you write once as structured YAML — boardwatch never parses a PDF or Word file:

boardwatch tailor init       # scaffold {config_dir}/resume.yaml, then edit it
boardwatch tailor validate   # confirm it loads: entry and bullet counts, detected skills
boardwatch run --top 10      # scan → eligibility → rank → liveness → tailored PDFs
boardwatch web               # review the delivered leads in your browser

Edit two files in your config directory before the first run (init and tailor init print their paths):

• resume.yaml — your skills and your experience entries and bullets: the only text a tailored résumé selects from. Its first header line is also your name in each PDF's file name and metadata, unless answers.yaml sets identity.full_name.
• resume_template.tex — the page layout, and the source of the name, contact line and education block printed at the top of the page. init wrote a starter copy, and a run refuses it until its placeholder identity is replaced ("Your Name", "555 555 5555", "you@example.com", "Example University", "Example Field"). Replace the rest of that block too — the city, the sample links, courses and GPA — because only those five phrases are checked.

What a run leaves behind:

• ~/boardwatch-queue/ — one folder per delivered lead, holding its tailored PDF, the job description, a link to the posting (apply.webloc on macOS, apply.url on Windows, apply-link.txt elsewhere) and a details.json. Leads a check could not clear wait in _review/, without a PDF, for you to read first; what you act on moves to _applied/, _skipped/ or _reported/.
• ~/boardwatch-applications/<date>/ — the day's render folders, a morning-<run>.md summary, and a funnel-<run>.md that accounts for every posting the run considered.

The review app reads that queue. It lists the apply and review lanes with each lead's PDF, job description and eligibility evidence, and marking a lead applied there keeps it from resurfacing, just like track. It binds 127.0.0.1 only and authenticates with a token kept in {config_dir}/web-token, so the URL it prints can be bookmarked. An optional {config_dir}/answers.yaml fills a copy-to-clipboard panel with the answers every application form asks for; nothing is ever typed into an employer's page.

Schedule run once a day and it becomes an unattended driver: see the unattended run guide for launchd, cron and systemd recipes, the heartbeat URL, and what a normal day's exit code means.

How it works

   init ──▶ scan ──▶ top ──▶ show <id> ──▶ track
   │        │        │        │             │
   companies fetch    rank on   full posting + record what
   + profile boards   demand    score breakdown you did

• init: one-time setup. Choose companies (starter set / registry search / paste), then your profile (résumé text, target titles, excludes, locations, remote-only).
• scan: fetches every watched board through a polite fetcher (per-host pacing, retries with backoff, conditional If-None-Match/If-Modified-Since so unchanged boards cost a 304), then applies each board transactionally. Prints a one-line summary.
• top [N]: ranks open postings against your profile right now (weights are read live), newest-and-most-relevant first, with a one-line "why".
• show <id>: the full posting plus a per-component score table (skill coverage, title match, recency, location fit) and the eligibility audit with quoted evidence.
• track: record what you did with a posting — add, status, list, log — so a lead you already applied to doesn't resurface tomorrow. See Your funnel.
• eligibility: facts / policy to tell boardwatch your situation, run to evaluate open postings, summary for a funnel of what the catalog matched. See Eligibility audit.
• companies: add / remove / search / list / import / export your watched boards. boardwatch companies add https://boards.greenhouse.io/acme just works.
• doctor: per-board connectivity and freshness, plus a local DB integrity check.
• config show / config set: tune politeness and ranking weights (see below).
• run and web: the whole pipeline as one scheduled command, and the local review page for what it delivered. See The daily pipeline.
• guide / skill: the full command reference and a short agent skill file, both generated from your installed copy.

Ranking, briefly

Each posting gets a 0–1 score: a weighted blend of skill coverage, title match (fuzzy), recency (exponential decay), and location fit. Undefined components renormalize away, so a sparse profile still ranks sensibly. Nothing is precomputed: change a weight and the next top reflects it. show <id> prints the exact arithmetic.

Eligibility audit

Ranking tells you how well a posting fits. The eligibility audit tells you whether you could apply at all, and shows its work.

You describe your situation once, in the catalog's own vocabulary:

boardwatch eligibility facts set work_authorization.status citizen
boardwatch eligibility facts set highest_degree bachelor
boardwatch eligibility facts set field_of_study computer_science
boardwatch eligibility facts set employment_type_preference fte_only
boardwatch eligibility policy set work_auth blocker   # treat this family as disqualifying

Then boardwatch eligibility run reads each open posting for catalogued requirements (visa sponsorship, security clearance, degree, years of experience, employment type, internships, location, and more), resolves them against your facts, and stores a verdict. boardwatch show <id> prints it with the receipts:

Eligibility: ineligible
  unmet · required: work authorization / sponsorship
      quote: "This role is not able to sponsor employment visas now or in the future."

Three properties are deliberate:

• Evidence-linked. Every requirement carries the exact sentence it was read from, sliced from the posting version that was evaluated. Nothing is paraphrased or invented.
• Deterministic by default. The same posting and the same facts always produce the same verdict. Out of the box there is no language model in the loop and nothing to hallucinate: the vocabulary and rules are a versioned catalog, and a verdict is invalidated and recomputed only when your profile or the catalog changes. There is an opt-in, off-by-default model-assisted final gate inside run (gate.enabled, which calls your own claude CLI): it can withhold a lead it reads as ineligible, quoting the sentence it relied on, and when it reads a posting as eligible it can release into the apply lane a lead held for review only because the rules found no requirements to read or could not decide its experience bar. Turning it off stops new judge calls; the readings it already recorded keep applying to ranking and to the lanes.
• Honest. A clean posting reads as "no flags", never as a guarantee. "no flags" means only that no catalogued disqualifier was found, not that you are cleared to apply, and ambiguous wording reads as "check" rather than a false all-clear.

boardwatch eligibility summary shows the funnel (how many postings were evaluated, the verdict split, and what fired per family) so you can watch the catalog working before you trust a hidden count. By default top hides postings that are ruled out and reports the count; top --include-ineligible shows them.

top also hides postings whose title is not a software role — a "Deal Strategist" or an "Asset Tracking Technician" that fuzzy title matching would otherwise float to the top. The same rule applies to notify. This filter is deliberately loud rather than silent: the count appears under the table, stats reports it, show <id> tells you what the gate made of any posting, and top --include-non-swe lists the hidden rows with the exact title text that vetoed each one. A title that gives no signal either way is never filtered — it is scored normally, so a genuine software job with an unusual title is not at risk.

top can also hide postings whose title names a seniority band above the one you target — a "Distinguished Engineer" or a "Vice President, Engineering" that would otherwise crowd out the roles you can actually get. This is off until you ask for it: the profile field target_seniority_band is one of entry, mid, senior or any, it defaults to any, and on any the gate short-circuits before reading a single title. Set it with boardwatch profile edit. Like the role gate it is loud: the count appears under the table, stats reports it, show <id> explains any posting, and top --include-over-seniority lists the hidden rows with the title text that decided each one.

Only a confident hit hides a posting; an unresolved level token abstains rather than guessing. Level-aware gating is optional and documented in configuration.

The application form is not part of the posting boardwatch reads, and on a Greenhouse board a form question stating a citizenship, US-person or export-control requirement holds the lead for you to read rather than writing a verdict — a form is not the posting an evidence chain can quote from. On every other board the form is invisible, so a requirement that lives only there is a permanent miss, and your own read of the form before you apply is the last gate.

A posting that has been taken down does not always answer as gone: a dead Ashby page answers HTTP 200 with an empty shell, and a dead Greenhouse page redirects to an error URL that also answers 200, so the per-URL liveness re-check reads both as alive. What closes them is absence from the company's own job list — seen in a scan of a board you watch, or through that ATS's list endpoint — so a stored URL with no board behind it can sit open indefinitely.

What changed since you last looked

boardwatch scan records every appearance, disappearance and body revision in an append-only ledger. digest reads the ledger from wherever you left off:

boardwatch digest          # new, reopened, updated, and a closed count
boardwatch digest --peek   # the same view without consuming it
boardwatch top --new       # rank only the postings first seen since your last digest

The cursor is an event id, not a timestamp, so a clock change or a missed day cannot skip or repeat a window.

Your funnel

boardwatch never applies for you. It records what you did, so the state stays yours:

boardwatch track add 42                    # start tracking a posting
boardwatch track status 1 applied          # move it, with an immutable ledger entry
boardwatch track status 1 interviewing --note "phone screen booked"
boardwatch track list --status applied
boardwatch track log 1                     # the full history for one application

Already applied to some of these through another tool? Import that history once and those roles stop re-surfacing. The file is a plain CSV or JSONL with the columns company, title, url, applied_at, status — whatever you are migrating from, you can produce it:

boardwatch track import history.csv --dry-run --report audit.jsonl
boardwatch track import history.csv --report audit.jsonl

A row is matched on its url, or — with --allow-title-match, which is weaker because one title at a large employer can cover several requisitions — on company and title. When that weaker key does cover several, the row is recorded against none of them: marking a job applied removes it from the queue for good, so guessing which requisition you meant is the one mistake you could not find afterwards. Nothing is dropped silently: every row is counted as matched, already present, unmatched, malformed or ambiguous, and --report writes each one out with the key that matched it and the jobs it covered. Re-running the same file writes nothing.

Where you stand

boardwatch stats is a one-screen, read-only readout over your local database — no network, no writes of its own:

boardwatch stats             # qualified opportunities (last 7 days) + the discovery pipeline
boardwatch stats --days 30   # widen the trailing window

The first view partitions recent postings into qualified / uncertain / ineligible / unevaluated; a posting you have not evaluated yet is unevaluated, never counted as qualified. The second view is the pipeline: seen → passes filters → not ineligible → tracked. It needs a profile, so run boardwatch init first.

Take your data with you

boardwatch export --format jsonl --out postings.jsonl
boardwatch export --format csv

Every row carries the posting, your funnel state, and the eligibility verdict together with the profile and rules hashes that identify the evaluation it was computed under. This is a flat snapshot, not a full audit trail; it does not support independent recomputation of verdicts.

Configuration

boardwatch config show prints every key, its value, and its default; boardwatch config set <key> <value> changes it (validated at set time and load time). The full key table lives in docs/configuration.md.

boardwatch settings gives a read-only view of every opt-in feature (LLM tiers, notifications) — state, what it does, what it sends anywhere — and boardwatch settings toggle flips them interactively; both share the same validation as config set. See docs/configuration.md.

Schedule scans and get notified — see scheduling.

Run the whole pipeline unattended — see the unattended run guide.

Tailor a résumé

boardwatch never parses a résumé — parsing free-form PDFs/Word docs to understand a person's history is exactly the kind of guessing this project avoids. Instead you author your résumé once as structured YAML, and tailor only ever renders it:

boardwatch tailor init                 # scaffold {config_dir}/resume.yaml, edit it in your editor
boardwatch tailor validate             # confirm it loads; see entry/bullet counts + detected skills
boardwatch tailor run <posting-id>     # tailor it against one posting's extracted JD skills

validate and run read {config_dir}/resume.yaml unless you pass --resume PATH. run also takes --out DIR (default {data_dir}/tailored), --format latex (the only adapter), and --dry-run (report only; writes no file and records no artifact). It prints one line per bullet — kept, reordered, swapped, or dropped, with the JD skills that bullet covers — and the same per-bullet audit is stored on the artifact row.

tailor run selects which of your authored bullets to keep (favoring ones that cover the posting's extracted skills), reorders skill mentions, and applies whole-token synonym swaps from a small, bundled, frozen equivalence table (e.g. "JS" ↔ "JavaScript") — never free text generation. Before anything is written, a no-fabrication guarantee re-checks the output against your master résumé: every kept bullet's tokens must be either unchanged or a substitution the equivalence table names, and no bullet, entry, or section can appear that wasn't in the original. A résumé that fails this check is rejected before any file or database row is written, never delivered as a "best effort".

Honest bounds (Tier A). This is Tier A: a local, deterministic bullet-selection and safe-synonym pass — it does not rewrite your prose, invent new claims, or call any model. Your profile text (the free-form blurb from boardwatch init) is never imported into the résumé — tailor reads only what you author in resume.yaml.

PDF output is a hard gate, not best-effort. The renderer is tectonic, compiling a LaTeX template at {config_dir}/resume_template.tex — required, not optional: a run refuses outright, naming the missing path, if that file is absent, rather than silently falling back to the bundled example template (whose header/education are placeholder text, e.g. "Your Name" / "you@example.com"). A copy of that bundled template that was never edited is refused too. A lead without a shippable PDF is refused rather than delivered as rendered source: a missing tectonic on PATH fails the run loudly, and a résumé that compiles to more pages than your profile's page limit (resume_max_pages, default 1, set with boardwatch profile edit) is rejected. Two binaries are required, not one — tectonic to compile and poppler's pdfinfo to count the pages; without pdfinfo the page-count gate cannot answer and every lead is refused. boardwatch doctor probes for both and exits non-zero if either is missing. Output lands at {data_dir}/tailored/<Your Name>_<Company>_<Role>.{tex,pdf} — named for the application, so it reads as itself in an employer's upload dialog, and the PDF's own Title and Author carry <Your Name> - <Role> - <Company> and your name. Your name comes from your answers.yaml identity.full_name or your résumé's header, never from the code. The path is still a deterministic function of the lead, so re-running tailor run for the same posting overwrites that file even though each run is recorded as its own artifact in the database; the file on disk always reflects your most recent run, not necessarily the one you're currently reading about. If a later compile fails, the stale PDF from the previous run is removed rather than left behind next to the new source.

Opt-in LLM rewriting is covered in tailoring.

Career-profile bundle (advanced, optional)

boardwatch profile-bundle is a private, revisioned, filesystem-only store for the career facts a résumé is assembled from: typed YAML records, evidence captured by digest, owner approval bound to a content digest, and immutable content-addressed revisions. It lives at {config_dir}/career-profile (override with --bundle PATH) and nothing leaves your machine.

It is an alternative to hand-writing resume.yaml, not a requirement. Projection turns a promoted revision into a résumé: profile-bundle approve-projection records your approval of the exact text it will print, and boardwatch run --project then renders each lead from the bundle instead of resume.yaml (resume project followed by tailor run --resume does the same for one posting). The bundle's on-disk format and JSON reports may still change between releases.

See docs/profile-bundle-authoring.md for the format, every command, the 0/1/2/3 exit contract, and recovery from a stale draft or a corrupt evidence blob, and docs/projection-rendering.md for the path from a promoted bundle to a PDF.

Supported boards

Naming a board. companies add takes a board URL or a provider:slug:

• Greenhouse, Lever, Ashby, Workable, SmartRecruiters — paste the board URL (https://boards.greenhouse.io/acme), or greenhouse:acme and so on.
• Workday needs three parts, not one. A Workday board is identified by a host, a tenant and a career-site slug, so its target form is workday:<host>/<tenant>/<CareerSite> — for example workday:acme.wd5.myworkdayjobs.com/acme/AcmeCareers. Pasting the career-site URL (acme.wd5.myworkdayjobs.com/AcmeCareers) works too and derives the tenant for you. Site slugs are case-sensitive; hosts and tenants are not. One tenant can serve several disjoint career sites, so each site is watched as its own board.
• Oracle HCM — paste the …/hcmUI/CandidateExperience/en/sites/<Site> URL, or oraclehcm:<host>/<Site>.
• Eightfold — paste https://<tenant>.eightfold.ai/careers, or eightfold:<host> for a career site on the employer's own domain.
• Jibe and Phenom live on the employer's own domain, so a pasted URL cannot name them: use jibe:<careers-host> and phenom:<host>/<country>/<lang>.
• Amazon and Apple are one board per job category or country: amazon:<category> (for example amazon:software-development) and apple:<country> (for example apple:united-states).

boardwatch ships a bundled registry of 37 verified public boards on Greenhouse, Lever, Ashby, Workable and SmartRecruiters, 15 of them a curated starter set, so init works offline out of the box. You can watch any board these providers host, not just the registry, with companies add. The registry is community-maintainable by PR; see src/boardwatch/registry/README.md.

Verifying a slug before you watch it. companies add and companies import are offline by default — they accept any well-formed provider:slug and let the next scan or doctor discover a typo. Pass --verify to probe each board first: boards that come back reachable are watched (a reachable-but-empty board is watched with a note), and boards that return 404 or cannot be reached are skipped rather than written. import --verify exits non-zero if it skipped anything, so a partial import does not read as a clean one.

Provider capabilities at a glance — bodies, registry counts, starter membership, and honest limits — are in the provider matrix; per-provider coverage limits are in provider notes.

Responsible use & legality

By default boardwatch reads the same public, keyless endpoints that power each company's own careers page — a JSON API for most providers, and for Apple (and the first request of an Eightfold scan) the data embedded in that public page. It does not log in, present credentials, or bypass any access control. That is deliberately the least-invasive way to get this data. Still, these are third-party services, and using them responsibly is on you:

• Keep the politeness defaults. The defaults (≥1 request/sec per host, conditional GETs, bounded retries, a descriptive User-Agent) are intentionally gentle. Don't crank scan_workers up or per_host_delay_seconds down to hammer a board.
• It's for personal job-search use, not bulk data resale or redistribution of posting content. boardwatch stores postings locally for your review.
• Provider terms & rate limits can change and may restrict automated access. You are responsible for complying with each provider's Terms of Service. If a provider asks you to stop, stop.
• No warranty. These are undocumented-stability public endpoints; they can change or break without notice.

If you're unsure whether your use is appropriate, err toward watching fewer boards, less often. A job seeker checking a dozen companies once a day is the intended shape.

The opt-in discovery lanes do not all meet that bar, which is why each is off unless you name it in lanes_enabled (linkedin, indeed, hiringcafe, jsonld, jobapps). They read listings beyond an employer's own board. jsonld — like the GitHub new-grad lists and grnh.se links that it and companies discover read — sends no key or credential and identifies itself honestly, and jobapps only reads a folder another local tool writes. The rest go further: the LinkedIn guest-search pages and the hiring.cafe search those lanes read are disallowed by those sites' robots.txt, and the Indeed lane calls Indeed's mobile-app API while presenting that app's own client identity. Each lane's module docstring under src/boardwatch/lanes/ states exactly what it requests and why. Whether to enable one is your decision, under that site's terms.

Privacy & data

• Local-first. Its primary store is one SQLite database in your platform data directory; the opt-in LLM tier also caches raw responses there as plain files on disk (override with --data-dir). No server, no account, no cloud.
• No telemetry. boardwatch phones home to nobody.
• Every credential you supply is opt-in and read from the environment only, never written to disk. The default path authenticates to nothing. The LLM tiers read BOARDWATCH_LLM_API_KEY, the webhook notifier BOARDWATCH_NOTIFY_WEBHOOK_URL, and an unattended run can ping BOARDWATCH_HEARTBEAT_URL (a dead-man's switch) and post its warnings to BOARDWATCH_ALERT_URL. The optional final-eligibility gate takes no key: it runs your own claude CLI. The one secret boardwatch creates itself is the review app's bearer token, written at mode 0600 to {config_dir}/web-token on first use. SECURITY.md says exactly what each one sends.

Where your files live.

Roadmap

• PyPI + GHCR published releases (pipx install boardwatch, docker run …)
• Notifications on new matches (desktop / webhook)
• digest and top --new change detection (only what changed since last run)
• More ATS providers: Workable, SmartRecruiters, Oracle HCM, Eightfold, Phenom, Jibe, and Amazon's and Apple's own career sites
• Data-portability export (--format jsonl|csv)
• Résumé tailoring (tailor init/validate/run, local, no-fabrication guarantee)
• Workday provider (host/tenant/site composite board identity)
• The unattended daily pipeline (boardwatch run) and a local review app (boardwatch web)
• More eligibility rule families (contract vs. full-time, internships)
• A readable settings surface, so every opt-in feature is discoverable and reversible without hand-editing config.toml
• Deduplication, so the same role posted twice is one lead and not two
• A durable decision ledger, so a lead you were already shown does not come back tomorrow
• Applied-state suppression, so a job you already applied to stays off the list
• A liveness re-check during run, so a requisition that has been taken down is dropped before a résumé is built for it

Next:

• An onboarding flow that fits fields other than software, so the eligibility taxonomy is gathered from you rather than assumed
• A measured acceptance run — the daily numbers published rather than asserted (in progress)
• Broader company coverage, but only after the above shows the funnel converts; breadth multiplies whatever is downstream of it, including the mistakes

These boxes track main, which can run ahead of the latest published release. CHANGELOG.md is the authoritative record of what shipped in the version pipx install boardwatch gives you; anything under Unreleased needs an install from source.

Have a company on a board boardwatch doesn't reach yet, or an ATS you want supported? Open an issue.

Contributing

Contributions welcome: code, registry entries, or bug reports. The smallest useful one is adding a public board — see the Contributing a board walkthrough, which has a one-command local check. See CONTRIBUTING.md for dev setup (uv sync, make check) and the registry guide for the catalog schema. All changes land via PR against a branch-protected main. Questions and ideas are welcome as issues too.

License

MIT.