gupy-search
Skill2 starsChanged 27 days agoMITView on GitHub
Use this skill whenever the user wants to search for jobs in Brazil, find Brazilian job listings, or look up a specific Gupy posting — even if they don't mention Gupy explicitly. Gupy is the applicant-tracking system behind the career pages of most large Brazilian employers (Itaú, Ambev, Boticário, Stefanini, Magalu, Natura, etc.), so its public portal is the broadest single source of corporate vacancies in the country. Covers CLT, PJ, estágio, trainee, jovem aprendiz, temporário and banco de talentos, filtered by state, city, workplace model (remoto/híbrido/presencial) and PcD/affirmative vacancies. Trigger phrases: vagas, vaga de emprego, buscar vagas, procurar emprego, emprego em São Paulo, vagas remotas, vagas home office, vagas CLT, vagas PJ, estágio, trainee, programa de trainee, jovem aprendiz, vagas PcD, gupy, jobs in Brazil, Brazilian jobs, work in Brazil, job search brazil, hiring brazil.
In this repository12 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- freehire-searchSkill
- jobbank-searchSkill
- jobdanmark-searchSkill
- jobindex-searchSkill
- jobnet-searchSkill
- linkedin-searchSkill
- vagas-searchSkill
- job-application-assistantSkill
- scrapeSkill
- upskillSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 29
Search Brazilian job openings by keyword, job title or technology
- Very specific: line 91
All errors go to stderr as { "error": "...", "code": "..." } with exit code 1.
- Very specific: line 95
Search uses the JSON endpoint behind portal.gupy.io's own search page.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Gupy Search Skill
Busca vagas ao vivo no Portal Gupy (portal.gupy.io). Sem autenticação, sem chave de API
e zero dependências de runtime — roda só com bun.
When to use this skill
- Search Brazilian job openings by keyword, job title or technology
- Narrow by state (UF), city, workplace model, contract regime (CLT, PJ, estágio, ...) or PcD
- Get the full posting (description, responsibilities, requirements, benefits, selection steps)
Commands
Search job listings
bun run .agents/skills/gupy-search/cli/src/cli.ts search --query "<texto>" [flags]Key flags:
--query <text>/-q <text>— keyword search. Required. Gupy matches the job title, so short title-like terms ("analista de dados", "desenvolvedor python") work better than long phrases.--state <UF|nome>/-s—SP,MG,"Rio Grande do Sul"... (UF is translated to the state name Gupy filters on).--city <nome>/-c— city exactly as Gupy stores it ("Belo Horizonte","Campinas").--remote <modo>—remoto|hibrido|presencial(Englishremote|hybrid|onsitealso accepted).--type <regime>—clt|pj|estagio|trainee|aprendiz|temporario|banco-de-talentos|freelancer|autonomo| ...--pcd— only vacancies flagged for people with disabilities (affirmative or open to PcD).--jobage <days>— posted within N days. Client-side filter: it implies--sort recent, and combining it with--sort relevanceis rejected.--sort recent|relevance— defaultrelevance(orrecentwhen--jobageis set).--page <n>— 1-indexed, 20 results per page.--limit <n>/-n— client-side cap.--format json|table|plain— defaultjson.
Remote postings have no city/state; their location is "Remoto". A --state/--city filter therefore excludes them — run a separate --remote remoto query for remote roles.
Fetch full job detail
bun run .agents/skills/gupy-search/cli/src/cli.ts detail <url> [--format json|plain]Pass the url from a search result (https://<empresa>.gupy.io/jobs/<id>); the tracking form
https://<empresa>.gupy.io/job/eyJ... is accepted too. A bare id is not enough — the
posting lives on the employer's subdomain. Returns description, responsibilities
(responsibilities), requirements (requirements), benefits/additional info
(additionalInfo), selection steps (steps), application cutoff (deadline) and isActive
(false = the posting is no longer published).
Usage examples
# Python, remote, last 14 days
bun run .agents/skills/gupy-search/cli/src/cli.ts search -q "python" --remote remoto --jobage 14 --format table
# Data analyst, CLT, São Paulo state
bun run .agents/skills/gupy-search/cli/src/cli.ts search -q "analista de dados" -s SP --type clt --format table
# Engineering internships in Minas Gerais
bun run .agents/skills/gupy-search/cli/src/cli.ts search -q "estágio engenharia" -s MG --type estagio --format table
# Full posting
bun run .agents/skills/gupy-search/cli/src/cli.ts detail https://empresa.gupy.io/jobs/12345678 --format plainOutput fields (search)
id, title, company, location ("Cidade / UF" or "Remoto"), workplace
(Remoto/Híbrido/Presencial), contractType (e.g. Efetivo (CLT), Pessoa Jurídica (PJ),
Estágio), date (published, YYYY-MM-DD), deadline (application cutoff), pcd, url.
meta.total is Gupy's match count (the endpoint caps some broad queries at 100).
All errors go to stderr as { "error": "...", "code": "..." } with exit code 1.
Notes
- Search uses the JSON endpoint behind portal.gupy.io's own search page; detail reads the
__NEXT_DATA__JSON embedded in the employer's career page.portal.gupy.io/robots.txtdisallows nothing. Keep volume low anyway — this is a personal job-search tool. - Gupy postings usually include steps such as Teste de perfil/Mapeamento comportamental,
Teste de lógica, Entrevista Inteligente (async video/AI screening) and Entrevista com o
gestor. The
stepsarray tells you which ones to prepare for — see.claude/skills/job-application-assistant/10-mercado-brasileiro.md. - Salary is almost never published on Gupy. Use
tools/clt_pj.pyand market research for benchmarks.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

