Troubleshooting
Version 1.2.0
Real issues that come up, and what's actually going on when they do.
"My content changes don't show up on the live site"
Check two things in order. First: is the content repository reachable from somewhere other than your own machine? The build container that produces the live site can only pull content from a real remote — a purely local content repository is completely valid for drafting, but the deployed site needs something to actually fetch from. Second: a deploy hook accepting the request only confirms the request was received, not that the build succeeded. Check the actual deployment status before assuming a publish went live.
"I got a git index-lock error"
This happens when two write operations overlap — most often an agent firing several tool calls in quick succession, not a multi-person conflict. It's expected, and it's why writes are serialized: retry the action, and it should go through cleanly the second time. If it happens repeatedly on the same action, that's worth reporting.
"I can't start a new draft on a post that's already being edited"
That's the lock working as intended, not a bug. A draft branch existing for a given entry is the lock — the error should tell you who holds it and since when. If it's a stale draft nobody's actually working on anymore, discard it before starting a new one.
"The connector can't reach the server, or my session shows no tools at all"
Access is checked live against your actual GitHub permissions on the content repository, every time you connect — not cached from a previous session. If you were recently added as a collaborator, or your permission level recently changed, reconnecting should pick up the current state. A session with zero tools available almost always means the connected identity doesn't have real collaborator access yet, not a broken connection.
"I deleted something by mistake"
Nothing is actually gone. Every version of every file still exists in git's own history regardless of what's currently live — restoring a deleted entry reads that history back out directly, not from a separate backup system.