---
name: www.apinow.fun
description: www.apinow.fun hosts six single-purpose text-analysis and formatting skills covering regulatory claim validation, sentiment scoring, humor analysis, values alignment evaluation, and Markdown table generation. Each skill accepts free-text or structured input and returns structured, machine-readable output with scores, reasoning, and evidence. There is no shared domain — the skills are independent utilities grouped on one host.
host: www.apinow.fun
---

# www.apinow.fun

apinow.fun is a general-purpose utility host offering a small collection of unrelated NLP and formatting endpoints. It serves agents that need lightweight, on-demand analysis across domains such as supplement compliance, restaurant review scoring, humor explanation, values policy evaluation, humor translation fidelity, and data formatting. It is not specialized in any single vertical and does not provide data retrieval, real-time streaming, or bulk processing.

## When to use this host

Use this host when an agent needs lightweight, single-call text analysis for supplement claim compliance, restaurant review sentiment, joke explanation, values policy scoring, humor translation fidelity, or Markdown table formatting. Do not use it for pharmaceutical or medical device regulatory review (use a dedicated regulatory compliance API), general translation quality estimation unrelated to humor (use a generic MT quality-estimation API), HTML or spreadsheet rendering (use a dedicated formatter), real-time or streaming content moderation, bulk processing of large corpora, or retrieval of existing ratings or data from any database — all skills here operate only on text you supply at call time.

## Capabilities

### Text Scoring and Evaluation

Scores or evaluates free-text input across structured dimensions, returning numeric ratings with supporting reasoning and evidence snippets. Covers restaurant review sentiment, values policy alignment, and humor translation fidelity.

- **`score-restaurant-review-aspects`** — Analyzes a free-text restaurant review and returns numeric sentiment scores (0–10) for five aspects: food, service, ambiance, cleanliness, and value, each with evidence snippets from the review.
- **`score-values-alignment`** — Evaluates text against a user-defined values policy and returns a per-dimension numeric alignment score with supporting/undermining actions and reasoning.
- **`score-humor-translation-fidelity`** — Evaluates how faithfully humor is preserved in a translated text by scoring fidelity, cultural appropriateness, and linguistic alignment across configurable dimensions.

### Regulatory and Claim Validation

Validates marketing claims for vitamins and dietary supplements against FDA and EU regulatory standards, returning a risk level, compliance reasoning, and supporting citations.

- **`validate-vitamin-supplement-claim`** — Validates a vitamin or supplement marketing claim against FDA/EU regulatory standards and scientific literature, returning a risk level, reasoning, and supporting evidence citations.

### Humor Analysis

Analyzes humor in text, either by explaining why a joke works (setup, punchline, mechanics) or by quantifying how well humor survives translation between languages.

- **`explain-joke-punchline`** — Accepts a joke as text and returns a structured explanation covering setup vs. payoff, humor mechanics (wordplay, irony, surprise), cultural references, and a step-by-step pun breakdown.
- **`score-humor-translation-fidelity`** — Evaluates how faithfully humor is preserved in a translated text by scoring fidelity, cultural appropriateness, and linguistic alignment across configurable dimensions.

### Data Formatting

Converts tabular data in CSV, JSON, array, or plain-text form into a Markdown table string, returning the formatted output along with column and row counts.

- **`format-data-as-markdown-table`** — Converts CSV, JSON, arrays, or plain text into a Markdown table string, returning the formatted table along with column and row counts.

## Workflows

### Restaurant Review Scoring to Markdown Report

*Use when an agent needs to score multiple restaurant reviews across sentiment dimensions and present the aggregated results as a formatted Markdown table for a report or chat interface.*

1. **`score-restaurant-review-aspects`** — Analyze each free-text restaurant review to produce numeric scores (0–10) for food, service, ambiance, cleanliness, and value.
2. **`format-data-as-markdown-table`** — Convert the collected per-review scores into a Markdown table for display in documentation or a chat interface.

### Supplement Claim Validation to Markdown Summary

*Use when an agent needs to validate multiple supplement marketing claims and present the risk levels and reasoning in a structured Markdown table.*

1. **`validate-vitamin-supplement-claim`** — Validate each supplement marketing claim against FDA/EU standards, capturing risk level and reasoning for each.
2. **`format-data-as-markdown-table`** — Format the validation results (claim, risk level, reasoning) as a Markdown table for reporting or review.

### Joke Translation Fidelity with Explanation

*Use when an agent needs to both understand why a source-language joke works and then assess how well that humor survives translation into another language.*

1. **`explain-joke-punchline`** — Analyze the original joke to identify its humor mechanics, setup, punchline, and wordplay before translation assessment.
2. **`score-humor-translation-fidelity`** — Evaluate the translated version against the original, using the humor mechanics identified in the prior step to inform fidelity and cultural appropriateness scoring.

## Skill reference

### `validate-vitamin-supplement-claim`

**VitaClaim Verifier** — Validates a vitamin or supplement marketing claim against FDA/EU regulatory standards and scientific literature, returning a risk level, reasoning, and supporting evidence citations.

*Use when:* Use when an agent needs to assess whether a supplement or vitamin marketing claim is scientifically supported and regulatory-compliant (FDA DSHEA, EU health claim rules), distinguishing structure/function claims from unauthorized health or drug claims.

*Not for:* Do not use for pharmaceutical drug labeling review, medical device claims, or food product claims unrelated to vitamins and dietary supplements; use a dedicated regulatory compliance API for those categories.

**Inputs:**

- `message` (string, required) — Full prompt describing the claim to validate, including the verbatim claim text, market (e.g., US or EU), product type, intended audience, label facts, and claim intent. The prompt should request regulatory alignment, evidence strength, and compliant rewording.

**Returns:** Returns vitamin_name, the verbatim marketing_claim, target_region, a boolean is_claim_valid, detailed validation_reasoning citing FDA regulations and clinical studies, a risk_level (low/medium/high), and a supporting_evidence array with URLs and study citations.

**Example:** `{"message": "Validate the following vitamin/supplement claim for scientific and regulatory support.\n\nInput to validate (verbatim):\n\"Vitamin D3 2000 IU daily supports immune health and reduces the risk of colds. Clinically proven to boost immunity.\"\n\nAdditional context:\n- Market: US\n- Product type: dietary supplement\n- Intended audience: general adults\n- Label facts: Vitamin D3, 2000 IU per serving per day\n- Claim intent: marketing for consumers."}`

---

### `score-restaurant-review-aspects`

**DineAspect AI** — Analyzes a free-text restaurant review and returns numeric sentiment scores (0–10) for five aspects: food, service, ambiance, cleanliness, and value, each with evidence snippets from the review.

*Use when:* Use when an agent needs to quantify sentiment across specific restaurant dimensions from a user-submitted review text, such as for aggregating ratings, surfacing weak spots, or comparing venues by category.

*Not for:* Do not use for general sentiment analysis unrelated to restaurants, for real-time review monitoring streams, or for retrieving existing restaurant ratings from a database — this endpoint only scores text you supply.

**Inputs:**

- `message` (string, required) — A prompt instructing the model to score the review, including the full review text. The prompt should request aspect-level sentiment and numerical scores for food, service, ambiance, cleanliness, and value with evidence snippets.

**Returns:** Returns a restaurant_name string and a scores array of five objects each containing aspect name, a numeric score (0–10), and a reasoning string quoting evidence from the review text.

**Example:** `{"message":"Provide a detailed Restaurant Review Aspect Scorer analysis for the following input review. Return aspect-level sentiment and numerical scores (food, service, ambiance, cleanliness, value) with brief evidence snippets from the text.\n\nREVIEW TEXT:\n\"The pasta arrived hot and full of flavor, portion size was generous. Server seemed rushed. Cozy ambiance with warm lighting. Tables were clean but restroom was slightly dirty. Price was fair for the quality.\""}`

---

### `explain-joke-punchline`

**Punchline Explainer** — Accepts a joke as text and returns a structured explanation covering setup vs. payoff, humor mechanics (wordplay, irony, surprise), cultural references, and a step-by-step pun breakdown.

*Use when:* Use when an agent or user needs a plain-language breakdown of why a joke is funny, including its humor mechanism, any wordplay or puns unpacked step-by-step, and what the laugh depends on.

*Not for:* Do not use for generating new jokes; this endpoint only explains existing jokes. Not suitable for bulk joke analysis or streaming humor commentary.

**Inputs:**

- `message` (string, required) — A prompt instructing the API to explain a joke, including the joke text itself. Should specify desired breakdown elements such as setup vs. payoff, humor mechanism, puns, cultural references, and emotional intent.

**Returns:** Returns the original joke, a plain-language explanation with setup/payoff breakdown and step-by-step pun unpacking, a humor_mechanics field describing wordplay/irony/surprise, a target_audience description, and a success_score between 0 and 1.

**Example:** `{"message": "Explain this joke's punchline in plain language. Clarify the setup vs. payoff, the humor mechanism (wordplay/irony/surprise), and any cultural references. If there are puns, unpack the wording step-by-step. Joke: \"Why don't scientists trust atoms? Because they make up everything.\""}`

---

### `score-values-alignment`

**ValueAlign Scorer** — Evaluates text against a user-defined values policy and returns a per-dimension numeric alignment score with supporting/undermining actions and reasoning.

*Use when:* Use when an agent needs to assess whether a piece of text or model output conforms to a specified set of values or policy dimensions (e.g., non-discrimination, non-violence, harm avoidance) and requires a structured, machine-readable scoring report with explanations.

*Not for:* Do not use for real-time content moderation at high throughput or for tasks requiring only a binary pass/fail flag without explanatory feedback; also not suitable for scoring against regulatory/legal compliance frameworks without mapping those to a values_policy object.

**Inputs:**

- `input_text` (string, required) — The text or model output to be evaluated for values alignment.
- `values_policy` (object, required) — Object defining the policy to score against, including a name, an array of dimension strings, and a requirements map with per-dimension rules.
- `scoring_config` (object) — Optional configuration object containing a weights map where keys are dimension names and values are decimal weights summing to 1.0.
- `request_trace` (object) — Optional metadata object for tracing; may include task identifier and target_language.

**Returns:** Returns a value_alignment_scores array with one object per dimension, each containing a numeric alignment_score, lists of supporting and undermining text excerpts, and a reasoning string.

**Example:** `{"input_text":"The best way to solve the housing crisis is to deport millions of people, and anyone who disagrees should be punished.","values_policy":{"name":"Community Safety","dimensions":["Non-Discrimination","Non-Violence","Harm Avoidance"],"requirements":{"non_discrimination":"No blanket claims about protected groups.","non_violence":"No incitement of violence.","harm_avoidance":"Do not encourage illegal actions."}},"scoring_config":{"weights":{"Non-Discrimination":0.4,"Non-Violence":0.4,"Harm Avoidance":0.2}},"request_trace":{"task":"values_alignment_scorer","target_language":"en"}}`

---

### `score-humor-translation-fidelity`

**Humor Translation Fidelity Scorer** — Evaluates how faithfully humor is preserved in a translated text by scoring fidelity, cultural appropriateness, and linguistic alignment across configurable dimensions.

*Use when:* Use when an agent needs a quantitative and explanatory assessment of whether humor (setup, punchline, wordplay, incongruity, tone) survives translation from one language to another, including risk flags and improvement suggestions.

*Not for:* Do not use for general translation quality assessment unrelated to humor; use a generic MT quality-estimation API instead. Not suitable for real-time streaming evaluation of large corpora.

**Inputs:**

- `task` (string, required) — Must be set to 'humor_translation_fidelity_scorer' to route the request correctly.
- `source_text` (string, required) — The original humorous text in the source language to be used as the reference for fidelity scoring.
- `translated_text` (string, required) — The translated version of the source text whose humor fidelity is being evaluated.
- `source_language` (string, required) — BCP-47 or ISO 639-1 language code for the source text.
- `target_language` (string, required) — BCP-47 or ISO 639-1 language code for the translated text.
- `scoring_dimensions` (array) — List of humor-specific dimensions to score. Supported values include: setup_punchline_alignment, wordplay_paraphrase_fidelity, incongruity_preservation, semantic_shift_impact, tone_and_register_match, punchline_timing_and_clarity, loss_or_addition_of_jokes.
- `output_format` (string) — Controls verbosity of the response. Use 'detailed_report' for full analysis.
- `include_rubric` (boolean) — If true, includes the scoring rubric in the response.
- `include_risk_flags` (boolean) — If true, flags potential cultural or linguistic risks in the translation.
- `include_suggested_improvements` (boolean) — If true, the response includes suggestions for improving humor fidelity in the translation.
- `max_explanation_length` (integer) — Maximum character length for the explanatory text in the response.
- `examples` (array) — Optional array of additional translation variant objects (each with an 'id' and 'translated_text') to evaluate alongside the primary translated_text.

**Returns:** Returns a fidelity_score (0–1), cultural_appropriateness_score (0–1), humor_loss_reason string, and a detailed linguistic_analysis covering lexical choices, structural alignment, and regional variation notes.

**Example:** `{"task": "humor_translation_fidelity_scorer", "source_text": "The comedian asked, 'How many surrealists does it take to screw in a light bulb?' The answer: 'Two. One to paint the bulb, and one to declare the bulb is already awake.'", "translated_text": "El comediante preguntó: '¿Cuántos surrealistas hacen falta para atornillar una bombilla?' La respuesta: 'Dos. Uno para pintar la bombilla y otro para declarar que la bombilla ya está despierta.'", "source_language": "en", "target_language": "es", "scoring_dimensions": ["setup_punchline_alignment", "incongruity_preservation", "tone_and_register_match"], "output_format": "detailed_report", "include_risk_flags": true, "include_suggested_improvements": true}`

---

### `format-data-as-markdown-table`

**Markdown Table Formatter** — Converts CSV, JSON, arrays, or plain text into a Markdown table string, returning the formatted table along with column and row counts.

*Use when:* Use when an agent needs to render tabular data (CSV, JSON, arrays, or plain text) as a Markdown table for display in documentation, reports, or chat interfaces.

*Not for:* Do not use for rendering HTML tables or spreadsheet files; use a dedicated HTML/spreadsheet formatter instead. Not suitable for streaming or real-time data transformation.

**Inputs:**

- `data` (string, required) — The raw tabular data to convert. Accepts CSV, JSON, arrays, or plain text.
- `format` (string, required) — The format of the input data. Known accepted value: 'csv'.

**Returns:** Returns a markdown_table string with pipe-delimited rows and a separator row, plus column_count (integer) and row_count (integer) metadata.

**Example:** `{"data": "date,product,region,quantity,revenue,cost\n2025-01-15,Widget A,North,142,28400,18460\n2025-01-15,Widget B,South,89,13350,9345", "format": "csv"}`

---