fix mcp framing + kuzu fts; add systemd watcher service commands and … - #811

@Ali110Nia is attempting to deploy a commit to the shashankss1205's projects Team on Vercel.

A member of the Team first needs to authorize it.

@Ali110Nia We have a lot of merge conflicts on this because of the recent refactor we did. can you please rebase it and then make a PR?

👋 Thanks for contributing to CodeGraphContext! Since this PR was opened, main has moved forward significantly (we just shipped v0.5.2 and merged ~50 PRs), so this one now has merge conflicts. Could you please rebase onto the latest main and resolve the conflicts? Once it is conflict-free and CI is green, we will review and merge it. Really appreciate your work — thank you! 🙏

Triage update: there are several distinct things in here (MCP framing fix, Kùzu FTS, systemd watcher service) totaling +3k lines — too much to review as one unit, and the branch has drifted far from main. @Ali110Nia if you're still interested, please split it: the MCP framing fix in particular sounds like something we'd want as a small focused PR. 🙏

Thanks for this, and sorry for the delay. This branch bundles several independent changes — MCP framing fixes, Kùzu FTS, systemd watcher service commands, plus docs — across 40 files, and it now produces ~20 conflict regions against main (including cli_helpers.py, config_manager.py, README.md, and five docs files). That's past the point where I can resolve it without guessing at intent.

Two of the pieces are also likely stale now:

• MCP framing — main pinned mcp>=1.0.0,<2 and fixed SSE JSON-RPC body buffering in fix: buffer full request body before SSE JSON-RPC parsing (#1113) #1319; worth re-checking whether your framing fix still reproduces.
• Kùzu FTS — note that refactor(kuzu): Reflect KuzuDB deprecation and provide users with an exit route (migrating to another DB) #1302 is currently proposing to deprecate KùzuDB as a runtime backend entirely. That's still an open question, but it may affect whether the FTS work is worth rebasing.

The systemd watcher service commands look genuinely additive and least affected by any of this — if you're up for splitting the branch, that's the piece I'd start with, as a small PR against current main.

Let me know how you'd like to proceed. If you'd rather not carry it forward, no problem — say the word and I'll close it.

Pulled the backend-neutral part of this out into a small, focused PR: #1634. It's just the repo_path normalization fix (no Kùzu FTS, no MCP framing, no systemd changes), rebased cleanly on current main, so it should be easy to review independent of the rest of this branch's fate.