This is a great project, thanks for collecting and sharing all of these resources! I have hacked together similar things in the past and also referenced APIs.guru for a large swath of resources - however, as you probably have also found, it's not the most well-kept list when it comes to pruning inactive services.
For example, https://integrations.sh/getsandbox.com/ has already been identified (probably from an LLM pass on the web result for its domain) as "parked" on a domain sale page, but APIs.guru still provides the API specs, and thus is listed in your index.
I don't have a comprehensive solution in mind yet, but I was curious if you'd thought of a way to keep a more updated set. I hesitate to make automated requests on a scheduled or per-PR basis from a GitHub Action, but actually sending a GET to the service's known docs URL seems like the closest to a sure-fire way to do it. I would be happy to submit a PR that attempts to test & filter out unavailable services, but it's also prone to spurious errors or docs downtime. Dispatching a research agent to look for hints about service interruption or shut-downs could work as well, but there's unbounded token use involved there, which also doesn't seem viable long-term.
In any case, I'm just feeling out your appetite to attempt to improve upon the staleness situation.
This is a great project, thanks for collecting and sharing all of these resources! I have hacked together similar things in the past and also referenced APIs.guru for a large swath of resources - however, as you probably have also found, it's not the most well-kept list when it comes to pruning inactive services.
For example, https://integrations.sh/getsandbox.com/ has already been identified (probably from an LLM pass on the web result for its domain) as "parked" on a domain sale page, but APIs.guru still provides the API specs, and thus is listed in your index.
I don't have a comprehensive solution in mind yet, but I was curious if you'd thought of a way to keep a more updated set. I hesitate to make automated requests on a scheduled or per-PR basis from a GitHub Action, but actually sending a GET to the service's known docs URL seems like the closest to a sure-fire way to do it. I would be happy to submit a PR that attempts to test & filter out unavailable services, but it's also prone to spurious errors or docs downtime. Dispatching a research agent to look for hints about service interruption or shut-downs could work as well, but there's unbounded token use involved there, which also doesn't seem viable long-term.
In any case, I'm just feeling out your appetite to attempt to improve upon the staleness situation.