get_hiringcafe_job_details
ChatGPTUse this when a user wants the current public details for one HiringCafe job ID or job-page URL slug. Returns the public description, requirements, compensation when available, and a HiringCafe job-page link if the listing is active. It does not access account data or submit an application.
get_hiringcafe_profile
ChatGPTUse this when the connected HiringCafe account needs to be identified or labeled. Returns only the profile represented by the current connection.
get_hiringcafe_resumes
ChatGPTUse this when a user asks for jobs matching a resume stored in their HiringCafe account. Returns bounded parsed summaries only; it never returns resume files, storage paths, public file URLs, or raw extracted text.
list_hiringcafe_saved_jobs
ChatGPTUse this when a user asks about jobs currently saved in their connected HiringCafe account. Returns up to 50 most recently saved jobs with bounded public fields, canonical HiringCafe links when available, and the HiringCafe tracker URL. It does not return applications, hidden jobs, stage history, or notes, and it does not modify the tracker.
list_hiringcafe_saved_searches
ChatGPTUse this when a user asks about searches saved in their HiringCafe account or wants to find current jobs using one of those saved criteria. It returns search names, filters, and HiringCafe result links; it does not modify saved searches.
refine_hiringcafe_job_results
ChatGPTUse this when the HiringCafe results UI submits a new keyword search or edited filter selection. Returns up to 20 hydrated public listings for the existing fullscreen results view. It does not access accounts, save searches, or submit applications.
render_hiringcafe_job_results
ChatGPTUse this when a HiringCafe job search has returned one or more jobs and those results need to be displayed. Pass every returned HiringCafe job ID once, in the original search order, plus the search summary, full-results URL, suggested filters, and search request from the search response. Renders the approved fullscreen HiringCafe card grid with interactive search and filters, and returns a compact main-chat fallback with direct job links, a full-results link, and recommended refinements. It does not access accounts or submit applications.
search_hiringcafe_jobs
ChatGPTUse this when a user wants to find, browse, narrow, compare, or shortlist active public job listings. It is also appropriate when the user asks to retry or refine a job search because previous results were too broad, too narrow, stale, or irrelevant. Translate the user's request into the available role, location, workplace arrangement, employment type, compensation, experience, education, and company filters. Returns 20 matching listings by default, with one result per company, for the dedicated HiringCafe results display. Do not invoke solely because the user expresses frustration; use it only when the user wants current job listings or help refining a listing search. Do not use it for general career advice, resume or cover-letter writing, interview preparation, private HiringCafe account data, or submitting applications.
get_job_details
ClaudeRetrieve the full posting for one HiringCafe job: complete description, requirements, compensation, and the application link. Call this after search_jobs when the user wants to know more about a specific role, or to tailor a resume or cover letter to it. Rules: 1. Render the job title as a markdown link to the returned URL. 2. Keep URLs intact — do not strip query parameters.
get_profile
ClaudeIdentify whose HiringCafe account is connected to this conversation. Call this when you need to name or label the account, not as a precondition for the other account tools — each of those checks its own access. Output: the stable opaque profile id, and the name and email when the account has them.
get_resumes
ClaudeRead the parsed summary of the resumes on this person's HiringCafe account. Call this when they ask for jobs matching their background, or to tailor an application to their own history. Output: job history, education, skills and certifications, bounded and parsed. The resume file itself, its storage path, its URL and its raw text are all excluded. When searching afterwards, the criteria relevant to what was asked give better results than everything the resume contains.
list_saved_searches
ClaudeList the searches this person has saved on HiringCafe. Call this when they refer to their own saved searches, or want current jobs for criteria they saved earlier. Output: each search's name, its filters in readable form, and a link to run it on HiringCafe. To return live results, read the filters and call search_jobs with the ones the person actually asked about — this tool returns saved criteria, not postings. It only reads them.
search_jobs
ClaudeSearch live job postings on HiringCafe. HiringCafe indexes millions of jobs scraped directly from employers' own applicant tracking systems, so listings link to the real application page rather than an aggregator reposting. Beyond keywords, this tool filters on structured fields that most job boards do not expose: seniority level, individual-contributor vs people-manager, employment type, posted-salary-only, and years of experience. It also filters on the employer itself — headcount, funding stage, investors, industry, organization type — which keyword search cannot reach. Do not tell the user these cannot be filtered, and do not fall back to your own knowledge of which companies are small or venture-backed: pass the arguments below and let the index answer. Composing common requests: • "startup" → max_employees (e.g. 500) with organization_types: ["private"]; add funding_stages for a specific stage. • "pre-IPO" → organization_types: ["private"] plus exclude_funding_stages: ["Post-IPO Equity", "Post-IPO Debt"]. There is no "pre-IPO" value in the data; it is private and not yet public. • "venture-backed" → funding_stages with the Seed/Series entries. • "big company" → min_employees. Rules: 1. Each result includes a HiringCafe link. Render the job title as a markdown link to that URL so the user can click through to apply. 2. Keep URLs intact — do not strip query parameters. 3. Prefer structured arguments over stuffing constraints into search. For example pass seniority_levels:["Entry Level"] rather than searching "entry level engineer".