PER-HOP DNS ANTI-SSRF GUARD
Zero Internal Network Exposure
Every URL and redirect hop (up to 5 hops) is resolved via DNS and verified against private IPv4/IPv6 ranges (127.0.0.0/8, 10.0.0.0/8, 169.254.0.0/16, ::1) before any socket opens.
v1.5.0 · MCP + API
Give Your AI Agents Grounded Web Data Without Token Bloat.
Ristria unifies real-time multi-provider web retrieval, hybrid ranking with recency and entity-overlap boosts, deterministic token-budget packing (max_tokens), serverless PDF extraction, and JS SPA crawling into one API, MCP server, and zero-token CLI.
- Response Time
- 2 ms
- Shared SerpCache
- HIT
- Packed Tokens
- 342 / 600
Example response. Press Execute to run it live, no key needed.
- https://docs.python.org/3/library/asyncio-task.html
The first time any of the tasks belonging to the group fails with an exception other than asyncio.CancelledError, the remaining tasks are cancelled.
MCP STDIO/Claude Code/REMOTE HTTP/Cursor/CLI/OpenCode/REST/Windsurf/
100k searches with highlights: $49 on Ristria vs $399–$1,600 at competitor list prices.
5/5 vs 0/5
Precision@5 vs Tavily on benchmark query s03 (Italian technical docs, Oct 2026)
2 ms
Shared SerpCache Hit
$0.49 / 1k
Team Plan Rate (vs $8–$16)
1 credit
= 1 search + BM25 highlights
Core endpoints
Web Search, Hybrid Ranking & Extraction Infrastructure
POST /v1/search // growth_search_cascade
1 credit (BM25 highlights included)1,078 ms cold // 2 ms warm
REQUEST // SEARCH
curl -X POST https://api.ristria.com/v1/search \
-H "Authorization: Bearer rs_free_YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "Python asyncio TaskGroup exception handling",
"category": "general",
"include_highlights": true,
"max_tokens": 600
}'[ .JSON ] RESPONSE200 OK
{
"query": "Python asyncio TaskGroup exception handling",
"total": 4,
"cache_hit": true,
"used_tokens": 342,
"response_time": 0.002,
"results": [
{
"position": 1,
"title": "Coroutines and Tasks — Python 3.14 documentation",
"url": "https://docs.python.org/3/library/asyncio-task.html",
"score": 0.7225,
"provider": "cascade",
"highlights": [
"The first time any of the tasks belonging to the group fails with an exception other than asyncio.CancelledError, the remaining tasks are cancelled."
]
}
],
"usage": { "credits_used": 1, "remaining_credits": 2499, "plan": "free" }
}Agent onboarding
Connect Any AI Agentin 10 Seconds
Whether your team builds with MCP clients, bash-first coding agents that avoid system prompt bloat, or LangChain and plain REST integrations, Ristria connects with a single credential.
4 ACCESS CHANNELS FROM 1 ENGINE
MCP Server (Claude Code, Cursor, OpenCode, Windsurf)
01 / 03
{
"mcpServers": {
"ristria": {
"command": "npx",
"args": ["-y", "@ristria/mcp"],
"env": {
"RISTRIA_API_KEY": "rs_free_YOUR_API_KEY",
"RISTRIA_API_URL": "https://api.ristria.com"
}
}
}
}Layered pipeline
Four Engineered Layers Under One Endpoint
Every request passes through four deterministic layers designed to minimize latency, protect context windows, and eliminate upstream search blocks.
ZERO EXTERNAL BINARY DEPENDENCIES
L02 ms warm hit
Shared 2-Tier SerpCache (RAM + SHA-256 Disk)
Normalizes whitespace, casing, locale, category, and domain filters into a canonical SHA-256 key. Repeat agent queries return in 2 ms at $0 upstream cost.
L1747–944 ms SERP
Multi-Provider SERP Fast-Path
Routes each query across multiple web and news search providers with freshness windows and automatic failover, so a single upstream outage never blocks your agents.
L2max_tokens exact fit
Hybrid Ranking & Token Budget Packing
Combines URL-aware Okapi BM25, rank fusion, recency and proof-count boosts, entity-overlap scoring, and best-first token budget packing.
L3247 ms extract
6-Tier Content, Serverless PDF & SPA Extractor
Extracts clean Markdown from HTML, /llms.txt, PDFs, and JS-heavy SPAs (headless rendering) with per-hop DNS anti-SSRF validation.
Instead of returning raw unranked HTML or charging 2x credits for passage extraction, Ristria ranks chunks with Okapi BM25 + hybrid recency and entity boosts and packs them into your model's exact token budget.
Fast, Token-Efficient, and Measured Against Tavily
VIEW FULL EVALUATION SUITE| CAPABILITY / METRIC | RISTRIA v1.5 | TAVILY | EXA AI | FIRECRAWL |
|---|---|---|---|---|
| 100k Searches + HighlightsList price per month | $49 / mo | $1,000–$1,600 | $700 | $399 |
| Italian Technical Docs P@5 (s03)Single benchmark query, Oct 2026 run | 5 / 5 | 0 / 5 | Not tested | Not tested |
Competitor prices are public list prices verified on 2026-10-14 from Tavily, Exa and Firecrawl (100k searches with extracted passages; Tavily advanced search and Firecrawl search bill 2 credits each). Precision@5 was measured on Ristria and Tavily only; methodology and raw results are available on request.
Built for enterprise agent fleets
Security, Governance & Production Safeguards
SHA-256 KEY HASHING & ZERO RETENTION
Cryptographic Key Metering
API keys are hashed immediately with SHA-256 in CreditMeter and never stored in plaintext. Queries are never used to train models.
MCP CONNECTOR + MANAGED CLOUD
Standard Interfaces, No Lock-In
One API key works across MCP, REST and the zero-token CLI, so moving in or out of Ristria is a one-line configuration change.