Find a prompt for anything.
Search the public registry in plain English, or browse trending prompts and curated collections from across the community.
Browse the registry
46 prompts · 0 collectionsGrade a source/translation pair on independent accuracy and fluency axes (0–1) with an itemized, severity-tagged issue list — usable directly as an LLM judge in evals.
Translate text with an explicit register knob — formal, informal, marketing, or technical (Sie/du, keigo, copywriter idiom) — plus notes on the tone choices made.
Adapt text for a target locale beyond literal translation — idioms, cultural references, units, date/currency formats, spelling — with an auditable list of every adaptation made.
Classify a message as ham, spam, or phishing (closed enum) with a confidence and the concrete signals behind the verdict — phishing wins when a message is both.
Moderate text against a closed policy taxonomy (harassment, hate, sexual, violence, self-harm, spam) with a safe verdict, severity-tagged flags, verbatim quotes, and a flat categories array.
Classify the emotion(s) a text expresses from a closed taxonomy (joy, anger, sadness, fear, surprise, disgust, neutral) with a primary emotion, secondary emotions, and intensity.
Classify a user message against a runtime intent taxonomy (label + description) for NLU and chatbots, with confidence and an explicit out-of-scope fallback flag.
Grade whether a summary is grounded in its source document (0–1) with each unsupported claim itemized — usable directly as an LLM judge in evals and as a runtime hallucination gate.
Generate a headline plus alternatives with a style knob (neutral / engaging / seo) and a hard word limit — every claim grounded in the text, no invented specifics.
Extract up to N key points from a document, each backed by a verbatim supporting quote — groundedness you can verify with a substring check, no judge needed.
Summarize an email thread into a brief with a closed status enum (resolved / waiting_reply / action_needed / informational), who the ball is with, and stated next steps.
Turn a meeting transcript into structured minutes — summary, decisions, action items, open questions — with owners and due dates only when explicitly stated, never inferred.