Karve

List Search Filters

Returns the valid skills and domains that search_jobs accepts as filters, plus the allowed remote_type values and the known location cities. Call this once before search_jobs so you map the user's request onto keys that actually exist in the data, instead of guessing free-text strings that won't match.

Search Jobs

Search and filter open jobs based on matching skills, domains, working arrangement, recency, or location. `remote_type` is one of 'remote', 'hybrid', 'onsite' — omit it to include all. Call the `list_search_filters` tool first to confirm you're using keys that actually exist in the data. Location: `locations` is a list of UK city names (e.g. ["London", "Bristol"]). A job matches if any of its offices equals a selected city or falls within `radius_miles` of one. `radi

Search Jobs Scored

Search and filter open jobs using the same filters as `search_jobs` (skills, domains, remote_type, recency, locations/radius_miles), then score each match against a stored profile's derived skills/domains/responsibilities using the same embedding-based relevancy scoring the chat UI uses (cosine similarity over cached OpenAI embeddings) — not the keyword-overlap count `search_jobs` uses. Call `list_search_filters` first to confirm valid skill/domain keys. Results are

Score Cv Against Job

Score a single candidate CV against a job description, 0-100, using the CV Screener's rubric-then-sum scoring (auto-generated question set, weighted by relevance to the JD). Returns the score, tier, per-question breakdown, and pros/cons. This is a one-shot convenience wrapper around the full CV Screener flow (which normally supports a batch of CVs and a recruiter-editable, forkable question set) for single-CV agent use. Requires a jobs_board MCP API key (tier 'pro' or hi