Reliability
What the last run measured.
We publish the numbers because most tools in this category do not. Every figure below comes from running 34 agent scenarios against the deployed servers, including the failure cases.
100%scenarios passed
34/34scenario results
4.10sslowest p95 across servers
2 Augmeasured 15:33 UTC
verifydesk
| Tool | Scenarios | Passed | p95 |
|---|---|---|---|
| verify_company | 4 | 100% | 2.69s |
| verify_vat | 2 | 100% | 3.28s |
| validate_iban | 4 | 100% | 165ms |
| screen_sanctions | 3 | 100% | 608ms |
| enrich_company | 2 | 100% | 1.23s |
| get_usage | 1 | 100% | 158ms |
jobsradar
| Tool | Scenarios | Passed | p95 |
|---|---|---|---|
| list_supported_companies | 2 | 100% | 757ms |
| search_jobs | 3 | 100% | 286ms |
| get_company_jobs | 3 | 100% | 4.10s |
| get_usage | 1 | 100% | 214ms |
invoiceforge
| Tool | Scenarios | Passed | p95 |
|---|---|---|---|
| generate_invoice | 3 | 100% | 569ms |
| validate_invoice | 2 | 100% | 172ms |
| extract_invoice | 2 | 100% | 172ms |
| describe_coverage | 1 | 100% | 171ms |
| get_usage | 1 | 100% | 174ms |
How this is measured
- The suite runs real calls against the deployed endpoint, not a mock: registry lookups, VAT checks, IBAN validation, sanctions screening.
- A scenario passes only if the answer is usable by an agent: right shape, no leaked internals, and errors that say what to do next.
- Failure cases count. Unknown companies, malformed input and unsupported countries are part of the run, because that is where tools usually break.
- p95 is the slowest response in the top 5 percent, measured end to end from outside Cloudflare, so it includes the upstream registries and job boards. Searching many job boards at once means the slowest board sets the number, and it moves between runs.
- Some sources throttle or go down without warning: the European VAT service does it regularly. When that happens the run is counted as degraded upstream, not as a pass, and not as our failure either, provided the tool returned a clean retryable error instead of breaking.
This page shows the most recent published run rather than a rolling uptime average. Continuous monitoring lands with the public launch, and the number here will move with it.