Search & Discovery
Find the right service for any task using natural language or structured WayforthQL queries. Results are ranked by WRI — Wayforth's live reliability index.
wayforth_search()
Natural language search across the Wayforth catalog. Returns services ranked by semantic relevance × WRI score.
Parameters
us-east, us-west, eu-west, ap-east. optional
card, usdc, x402. optional
10, max 50. optional
Return type: ServiceResult
wayforth_execute(slug=...)
translation, vision)
Examples
# Basic search
wayforth_search("image classification")
# With filters
wayforth_search(
"transcribe audio to text",
max_price_usd=0.05,
max_latency_ms=3000,
region="us-east"
)
# By payment rail
wayforth_search("generate embeddings", payment_rail="x402")
# Top 3 results
wayforth_search("code review", tier=3, limit=3)
WayforthQL v1.1
WayforthQL is a structured query language for precise service discovery. Use it when you need exact filtering, complex conditions, or reproducible queries.
Syntax
# Field filter
category:translation
# Comparison operators: :, :>, :>=, :<, :<=, :!=
tier:>=2
max_price_usd:<=0.05
# Boolean operators (AND, OR, NOT)
category:nlp AND tier:>=2
category:translation OR category:nlp
NOT category:storage
# Sort (SORT BY field ASC|DESC)
category:vision SORT BY wri_score DESC
# Limit
category:audio LIMIT 5
# Slug lookup
slug:deepl
# Full example
category:translation AND tier:>=2 AND region:eu-west SORT BY wri_score DESC LIMIT 10
Using WayforthQL via MCP
wayforth_query("category:translation AND tier:>=2 SORT BY wri_score DESC")
Via SDK
from wayforth import Wayforth
client = Wayforth(api_key="wf_live_...")
results = client.query("category:vision AND max_price_usd:<=0.01 SORT BY wri_score DESC LIMIT 5")
Field reference
| Field | Type | Description | Example |
|---|---|---|---|
category | string | Category slug | category:translation |
slug | string | Exact service slug | slug:deepl |
tier | int (1–5) | Service tier | tier:>=3 |
wri_score | int (0–100) | Reliability index | wri_score:>=80 |
max_price_usd | float | Cost per call | max_price_usd:<=0.05 |
max_latency_ms | int | P95 latency in ms | max_latency_ms:<=2000 |
region | string | Serving region | region:eu-west |
payment_rail | string | Payment method | payment_rail:x402 |
Tip: WayforthQL is case-insensitive for field names and operators. CATEGORY:Translation and category:translation are equivalent.
Search filters
All filters apply to both wayforth_search() (as kwargs) and WayforthQL (as field conditions).
| Filter | Type | Description |
|---|---|---|
tier |
int 1–5 | Service tier. 1 = community, 3 = production-ready (WRI ≥ 80), 5 = enterprise SLA. |
max_price_usd |
float | Maximum cost per successful call. Calls that fail before execution are not charged. |
max_latency_ms |
int | Maximum P95 round-trip latency in milliseconds, measured at the gateway. |
region |
string |
Preferred region:
us-east, us-west, eu-west, eu-central, ap-east, ap-south
|
payment_rail |
string |
Accepted payment method:
card (Stripe), usdc (Base L2), x402 (HTTP-native).
See Payment Rails.
|
limit |
int | Number of results to return. Default 10, max 50. |
Categories
All 19 categories available in the catalog. Use the category field in WayforthQL or browse with wayforth_services(category="...").
| Slug | Full name | Description |
|---|---|---|
translation | Translation & Localization | Machine translation, language detection, locale adaption |
nlp | Natural Language Processing | Named entity recognition, sentiment, POS tagging |
generation | AI Content Generation | Text, image, and structured content generation |
vision | Computer Vision | Image classification, object detection, OCR |
audio | Audio & Speech | Transcription, TTS, speaker diarization, voice cloning |
code | Code Intelligence | Code review, completion, vulnerability scanning, docgen |
embeddings | Embeddings & Vectors | Dense/sparse embeddings, semantic similarity, reranking |
search | Search & Retrieval | Web search, news retrieval, SERP, knowledge bases |
extraction | Data Extraction & Parsing | Document parsing, table extraction, HTML scraping |
summarization | Summarization | Long document summarization, abstractive & extractive |
classification | Classification & Tagging | Intent classification, tagging, zero-shot classification |
moderation | Content Moderation | Toxicity detection, NSFW filtering, policy enforcement |
finance | Financial Data | Market data, earnings, company filings, FX rates |
geo | Geolocation & Maps | Geocoding, IP geolocation, routing, place search |
identity | Identity & Verification | Email verification, phone lookup, ID document checks |
notifications | Notifications & Messaging | SMS, email, push, webhook delivery |
storage | File & Storage | Object storage, file conversion, CDN delivery |
analytics | Analytics & Intelligence | Web analytics, cohort analysis, funnel tracking |
data | Data Processing & ETL | Transformation, deduplication, enrichment pipelines |
WRI Score
The Wayforth Reliability Index (WRI) is a live 0–100 score that measures how trustworthy a service is for production use. It is recalculated continuously from real call data.
Score components
| Signal | Weight | Description |
|---|---|---|
| Uptime reliability | 35% | Rolling 30-day successful response rate |
| Payment success rate | 25% | Fraction of billing events that settled without retry |
| Latency consistency | 20% | P95 / P50 ratio — lower variance = higher score |
| Call volume | 15% | Log-scaled volume signal; higher usage = more trust |
| Tier multiplier | +5% | Tier 2+ services receive a reliability bonus |
Score thresholds
Services with WRI ≥ 80 are recommended for production workloads. The default search results are sorted by a weighted combination of semantic relevance and WRI score.
Note: WRI is computed for each service independently. A service can drop below 80 if its upstream provider degrades. Wayforth's self-healing payment layer will automatically retry or reroute calls to a backup provider when this happens.