The top IBAN ends in 32. The bottom one ends in 33. A length check passes both. The checksum catches the typo in the second one.
That’s what iban-validate can tell you. Checking who owns the account takes a separate step.
Say your agent already searches the web. Now it needs to look up a company. With Strale, you can add that lookup without setting up another provider account. You pay per call.
strale.io
MCP gives you a protocol for tool discovery. It does not give you a signal about which tools are reliable.
Thousands of MCP servers and growing. No quality scores. No test results. No way to know which ones break on edge cases.
That is a solvable problem.
We pull company data from 20 countries. Some via direct government APIs: Norway's Bronnoysund, France's api.gouv.fr, Finland's PRH, Switzerland's Zefix. Others via commercial registries and public data sources.
Every response tells you exactly where the data came from. The provenance field shows the source, method, and fetch time. No black boxes.
Every MCP server has a description. Almost none have a test suite.
Your agent picks tools based on the description alone. It has no signal about correctness, schema stability, or failure rate.
Quality metadata should be part of the MCP spec, not an afterthought.
Validate an email address: syntax, domain, MX records, disposable domain check, in one call:
curl -X POST api.strale.io/v1/do \n -d '{ "task": "email-validate", "inputs": { "email": "[email protected]" } }'
Free. No API key. Checks against 5,361 known disposable domains.
The EU AI Act requires logging for high-risk AI systems. If your agent is in scope, every data call needs a paper trail.
Every Strale API response includes: data source, fetch timestamp, quality score at execution time, provenance chain, data jurisdiction.
Built in from day one, not bolted on after the regulation.
Eight ways to connect your agent to Strale:
- Raw HTTP (any language)
- TypeScript SDK
- Python SDK
- LangChain plugin
- CrewAI plugin
- Semantic Kernel plugin
- MCP server (Claude, Cursor, Windsurf)
- x402 pay-per-use (Base mainnet)
Same 297 capabilities. Same quality scores. Same audit trails.
Pick the protocol your agent already speaks.
We score every capability on two dimensions:
Quality Profile: correctness, schema stability, error handling.
Reliability Profile: availability, success rate, upstream health.
These combine into SQS (0 to 100).
Current distribution: 86% score A-grade. The rest show you exactly what is degraded.
Full methodology deep-dive: dev.to/petter-strale/how-we-…
KYB verification for 20 European and global markets, from a single API.
SE, NO, DK, FI, UK, DE, FR, NL, BE, AT, IE, ES, IT, CH, PL, PT, US, CA, AU, SG.
Registry lookup, VAT validation, sanctions screening, LEI check. Bundled into one call per country.
Built in Sweden. Every response includes data jurisdiction and provenance metadata.
The hardest part of building agent data infrastructure is not the API layer. It is knowing when a data source has silently degraded.
A company registry that returns stale data looks exactly like one that returns fresh data. Unless you are testing continuously.
Your agent needs to verify a Swedish company before onboarding.
One API call. Four checks: registry lookup, VAT validation, sanctions screening, LEI verification.
Structured result. Audit trail. Quality scored.
EUR 1.50 per call. No subscription. 20 countries available.
Works with LangChain, CrewAI, MCP, or raw HTTP. Same API, same quality guarantees. strale.dev/docs
Every capability on Strale has a public quality endpoint.
GET /v1/quality/iban-validate
Returns the current SQS score, quality profile, reliability profile, and health state. Free. No auth.
Check any capability score before you rely on it.








