Upstream: fullsend-ai/issues/5686
We've had several times now where we make an improvement in the platform and documentation does not stay in sync. Some examples from the last couple months include the installation procedure, BYOA, and customizing agents. Lately the old docs have been deprecated 👍🏼, but the docs to match the current release have trailed by multiple weeks.
Today's example: 0.32 was released 8 days ago, the old method to customize an agent is deprecated and doesn't appear to work anymore, but docs haven't been updated with the replacement. As a user it's really difficult to understand how to use this critical feature.
On past teams we've resolved this through different means, e.g. a process change that PRs can't be merged w/o accompanying doc changes (if required), or a process change that features can't ship in a release unless the docs are complete (e.g. features must be documented, hidden behind an experimental flag, or removed before each release).
Another possible solution would be an agent that auto-updates the docs / opens PRs whenever docs are out of sync. Would be good to flag issues, but I'm not sure it would be able to auto-fix them/document new support. 🤷🏼♂️
Upstream: fullsend-ai/issues/5686
We've had several times now where we make an improvement in the platform and documentation does not stay in sync. Some examples from the last couple months include the installation procedure, BYOA, and customizing agents. Lately the old docs have been deprecated 👍🏼, but the docs to match the current release have trailed by multiple weeks.
Today's example: 0.32 was released 8 days ago, the old method to customize an agent is deprecated and doesn't appear to work anymore, but docs haven't been updated with the replacement. As a user it's really difficult to understand how to use this critical feature.
On past teams we've resolved this through different means, e.g. a process change that PRs can't be merged w/o accompanying doc changes (if required), or a process change that features can't ship in a release unless the docs are complete (e.g. features must be documented, hidden behind an experimental flag, or removed before each release).
Another possible solution would be an agent that auto-updates the docs / opens PRs whenever docs are out of sync. Would be good to flag issues, but I'm not sure it would be able to auto-fix them/document new support. 🤷🏼♂️