✅ #1 Use OpenAI SDK for Responses API streaming +3,383 -311 8d ago self by
mwcyu ↗ analysis ↗ GitHub Replace the hand-rolled fetch + SSE parsing in the OpenAI adapter with the
mwcyu/Mike-fork
mwcyu is hardening Mike's guts - test coverage, one unified AI pipeline, and live web search in the chat.
This is a fork by mwcyu, and so far the work is about making Mike sturdier and smarter under the hood rather than rebranding it into something new. There's no rename, no new niche, no separate product identity - it still reads as Mike, just with more discipline behind it.
Three threads tell the story. mwcyu wrote the automated back-end test suite the original never had, so the server-side wiring can't silently break. The AI plumbing that used to run as three near-identical pipelines now shares a single provider-agnostic driver. And the main chat can now reach out to the live web mid-conversation and fold current results into its answers.
Taken together, this is an engineer's fork: less polish-on-the-surface, more confidence-in-the-foundations, with one genuinely user-facing upgrade in the form of live search. If you care about how Mike is built - and whether it's built to last - it's worth a click through to GitHub.
What's in it
- Live web search in chat The assistant can reach the open web mid-conversation and pull current results straight into its answers.
- Unified AI driver Three near-identical model pipelines collapsed into one provider-agnostic path, so swapping or maintaining providers stops meaning triple the work.
- Back-end test suite Automated integration tests around the server-side plumbing - the safety net the original fork never bothered to build.
Direction
infrastructuresearchintegration
Activity
This fork is the first to wrap Mike's server-side plumbing in an automated test suite, so changes can't quietly break the wiring.
📝 mwcyu stops maintaining its AI plumbing in triplicate 3 commits 8d ago infrastructureintegration draft
This fork ran three near-identical pipelines to talk to its AI models - now they share one.
The assistant can now reach out to the open web mid-conversation and fold current results straight into its answer.
Threads of work (detailed view)
mwcyu writes the back-end tests upstream never bothered with
This fork is the first to wrap Mike's server-side plumbing in an automated test suite, so changes can't quietly break the wiring.
mwcyu stops maintaining its AI plumbing in triplicate
This fork ran three near-identical pipelines to talk to its AI models - now they share one.
mwcyu wires live web search into Mike's main chat
The assistant can now reach out to the open web mid-conversation and fold current results straight into its answer.
Pull requests (detailed view)
✅ Merged (1)
by
mwcyu · opened 8d ago · merged 8d ago by mwcyu