today
/today
Prowadzi cały dzień poszukiwania pracy w jednej sesji: synchronizuje Gmaila, ocenia oferty, przegląda CV, szkicuje odpowiedzi i wypełnia formularze do wysłania po zatwierdzeniu.
- Autor
- yusanglee ↗
- Repozytorium
- get-me-hired ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 11.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 2
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 11.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Prowadzi cały dzień poszukiwania pracy w jednej sesji: synchronizuje Gmaila, ocenia oferty, przegląda CV, szkicuje odpowiedzi i wypełnia formularze do wysłania po zatwierdzeniu.
Argumenty
/today [YYYY-MM-DD to pick up an earlier day, default today]
Oryginalny opis (EN)
Runs the whole job-search day in one session, from Gmail sync, finding and scoring roles, reads and picks through resumes, drafts, forms filled up to Submit, submit approvals, LinkedIn follow-up and wrap-up. Use to start each day (run today's job search) or to pick a day up again.
Jak działa
- The session's model
- Steps
- Telling the user how to start
Podgląd SKILL.md
---
name: today
description: Runs the whole job-search day in one session, from Gmail sync, finding and scoring roles, reads and picks through resumes, drafts, forms filled up to Submit, submit approvals, LinkedIn follow-up and wrap-up. Use to start each day (run today's job search) or to pick a day up again.
argument-hint: "[YYYY-MM-DD to pick up an earlier day, default today]"
model: opus
---
# Today: the whole day in one session
The one skill to start each day; "run today's job search" means the same. The whole day runs **in this session**:
every step that needs its own model runs as an agent inside it, and nothing starts another session. At the end the
user says "wrap up"; tomorrow they open a new session and start again. If the session closes or compacts mid-day, a
new one picks up from the files (`gmh today`).
Before the first step, read [rules.md](../../references/rules.md) (approvals, data, facts, no gaps, one session) and
[platforms.md](../../references/platforms.md) (finding `gmh` and the workspace `<ws>`, and every fallback). Agents
and their models: [agents.md](../../references/agents.md). How to ask: [questions.md](../../references/questions.md).
## The session's model
This session holds the approvals and the submits all day, so it should be on the strongest model (Opus in Claude
Code). On the first run, check which model it is on. If it isn't, say so once, in one line, and ask the user to
switch (`/model opus`) before going on. Don't refuse to work if they decline.
## Steps
`gmh today` ends with the suggested next step and its name (`[step: find]`); the names are in brackets below.
**Keep going between steps.** Don't end a turn or ask to start the next step: the Gmail sync starts while discover
runs, reads start when score finishes, the review page opens when the reads are applied, and prepare starts once the
picks are written. Stop only for what the user decides ([rules.md](../../references/rules.md): Gmail rows for the
tracker, the picks, the one `missing` question, new form answers, drafts with open checks, submits), when a step fails, or when the
user asks to stop. A failing `gmh` command is reported in one line with its error and the day goes on without that
step where it can; don't read engine source in this session to debug it.
0. **Where are we?** `gmh today --workspace <ws>` (add `--json` for the ids). Show its lines as printed. On a new
day, start step 1 right away. Ask first only when work is open from an earlier day (offer that queue or today's
run, in one question) or the user asked for something else (step 8). With a date argument, work that day's queue
(`--date <day>`) and skip steps 2 to 4.
1. **Check** [check]: [find-and-score.md](../../references/find-and-score.md), step 1.
2. **Find and score** [find, score]: [find-and-score.md](../../references/find-and-score.md), including the rerun
guard (two options; `--force` only on "Run it again anyway"). While discover runs, do the Gmail sync (step 2a);
then score, and go on to the reads.
2a. **Gmail sync at the start of the day**, whenever `gmh today` says it hasn't run today (also on a day with
nothing to find): [gmail.md](../../references/gmail.md), the reading done by the `gmail-reader` agent in the
background while discover runs. Every tracker write is a row the user
approved. If Gmail isn't connected, say so in one line and carry on with the day without it. An application the
sync adds (made outside get-me-hired) is caught by the applied-check in step 5 if today's run found it too.
3. **Reads** [reads]: [reads.md](../../references/reads.md), for every A, B and gap-tier role not read yet, started
as soon as score finishes. Then the review page, without asking.
4. **Review and pick** [pick]: [review-run](../review-run/SKILL.md), from the run summary through the picks (the
review page, always published to the user's Run Review artifact by the `review-publisher` agent where there is an
artifact tool, never the local file; then `gmh review picks` with a dry run first and the user's yes). In that same question, ask whether the
roles not picked should be passed; pass them only if the user says so. Then prepare, without asking.
5. **Prepare** [prepare]: [prepare.md](../../references/prepare.md), in its order:
a. the applied-check (stop on any role already applied to), the open-check, and reads for picks with none;
b. **resume updates for today's picks, before any resume is built** (`gmh materials candidates --date <day>`),
offered once;
c. `resume-tailor` per role and `gmh specs accept` (no per-role review); the one `missing` question if any; then
build every accepted resume (`gmh specs`, `gmh tailor`), show one line per role (company, template, PASS or
FAIL) and ask only about the failures;
d. forms, answers (new questions to the user once, batched); then `application-drafter` and
`application-reviewer` per role (no per-role text review: one line per role, ask only about open checks); then
`gmh notes` right away.
6. **Fill and submit in batches by tier, A first** [fill, submit]:
[fill-and-submit.md](../../references/fill-and-submit.md). `form-filler` fills one role at a time; one question
per batch names each role; only the ids the user names are submitted, from this session; `gmh applied` records
the confirmed ones. Manual roles: `gmh manual-notes`, then the user applies. An earlier day's queue [earlier]:
steps 5 and 6 with `--date <day>`.
7. **LinkedIn follow-up, optional** [linkedin]: when `gmh applied` says the day's queue is done, offer it once; also
whenever the user asks. Follow [linkedin-followup](../linkedin-followup/SKILL.md) in this session (its rules
hold: the user clicks and sends everything). The company loop stays here; people search is the
`linkedin-people-finder` agent; the note is the `linkedin-note-drafter` agent, given the path
`gmh linkedin packet <key> --workspace <ws>` prints, and its note is checked with `gmh linkedin table` as that
skill says.
8. **Anything else the user asks for today**, in this session, following that skill:
[gmail-sync](../gmail-sync/SKILL.md), [outcome](../outcome/SKILL.md), [followup](../followup/SKILL.md),
[interview](../interview/SKILL.md), the resume look ([design](../design/SKILL.md)) or resume updates
([materials](../materials/SKILL.md)). Their own summary step becomes part of the wrap-up.
9. **Wrap up** [wrap] when the user says so ("wrap up", "done for today"):
[wrap-up.md](references/wrap-up.md): a Gmail sync first, then the day's summary, `gmh home log today --auto`
and `gmh tracker report`.
## Telling the user how to start
In Claude Code the user types `/get-me-hired:today` or says "run today's job search"; other tools run the today
skill by name.
Pliki w paczce (2)
- SKILL.md 7 KB
- references/wrap-up.md 2 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · yusanglee/get-me-hired | 90 | MIT |
| victoranto/cycling-training | 100 | CC-BY-4.0 |
| talent-factory/claude-plugins | 100 | MIT |
| sejas/ai-code-skills | 100 | MIT |
| erayozenn/BigHead-2ndBrain | 90 | MIT |
| gkhurana21/co-op-coach | 86 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 0242dd5e3684 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100apply-all
Przechodzi przez dzienną kolejkę aplikacji: czyta ogłoszenia, przygotowuje materiały i wypełnia formularze partiami z jednym zatwierdzeniem. - 90daily-run
Szuka nowych ofert pracy, ocenia je punktowo, pokazuje stronę do przeglądu i przyjmuje wybory użytkownika. - 100job-batch
Prowadzi dzienną partię poszukiwania pracy: szuka ofert, ocenia je, przygotowuje CV i listy, aplikuje po zatwierdzeniu i aktualizuje tracker. - 90memory-today
Przypomina bieżący dzień w czasie lokalnym, gdy użytkownik pyta o dzisiejszą pracę lub ustalenia. - 100coordinate
Koordynuje sesję poszukiwania pracy: prowadzi ewidencję i decyzje, uruchamia codzienne szukanie i otwiera osobną sesję dla każdej oferty. - 100routine-morning
Prowadzi krótki start dnia w kilku krokach: domyka poprzedni dzień, zamienia skrzynkę i źródła na zadania i planuje dzień wokół kalendarza.