Re: Exploring MCP-based tools for Incubator insights

That's a fair point :)

Regards
JB

On Tue, Apr 21, 2026 at 4:54 PM Jarek Potiuk <[email protected]> wrote:

> > What do you think also about adding a CLI on top of the MCP
>
> Hmm. I just asked Claude to add the MCPs by giving it the URLs and it did
> it did it and you can work with it straight from Claude CLI :)
>
> J.
>
>
> On Tue, Apr 21, 2026 at 4:05 PM Jean-Baptiste Onofré <[email protected]>
> wrote:
>
> > Hi Justin
> >
> > That's pretty cool! Thanks!
> >
> > What do you think also about adding a CLI on top of the MCP. The reason I
> > think CLI is interesting is because it's easier to integrate with Claude
> > Code :)
> > Happy to work on something (as I have a prototype in progress in Apache
> > Karaf).
> >
> > Regards
> > JB
> >
> > On Sun, Apr 19, 2026 at 3:55 AM Justin Mclean <[email protected]>
> > wrote:
> >
> > > Hi,
> > >
> > > I have been experimenting with a small set of tools to explore how we
> > > might better understand Incubator data and podling signals in a more
> > > queryable way.
> > >
> > > The approach is based on simple MCP (Model Context Protocol) servers
> that
> > > expose different layers of Incubator information:
> > > - PodlingsMCP: lifecycle and metadata from podlings.xml
> > > - HealthMCP: parsed metrics from the incubator health reports
> > > - IncubatorMCP: combines the above to produce IPMC-oriented views such
> as
> > > watchlists, readiness signals, and podling briefs
> > >
> > > There is also a small skills file included to make it easier for MCP
> > > clients to use these tools.
> > >
> > > The intent is not to replace existing processes or judgment, but to
> make
> > > it easier to:
> > > -  identify podlings that may need attention
> > > - spot patterns across reports and activity
> > > - generate quick summaries for discussion
> > > - experiment with what the data suggests before applying human context
> > >
> > > The higher-level layer (IncubatorMCP) keeps a clear separation between:
> > > -  source facts (from podlings.xml and health reports)
> > > - derived signals (heuristics around activity, mentoring, reporting,
> and
> > > so on)
> > >
> > > This is still early and intentionally lightweight, but it may be useful
> > as:
> > > - a way to explore Incubator data programmatically
> > > - a potential aid for reporting or mentoring workflows
> > > - a foundation for further experimentation
> > >
> > > Repositories are here if anyone is interested:
> > > -  https://github.com/justinmclean/PodlingsMCP
> > > - https://github.com/justinmclean/HealthMCP
> > > - https://github.com/justinmclean/IncubatorMCP
> > >
> > > I would be interested in feedback on a few things:
> > > - whether this kind of tooling would be useful for IPMC work
> > > - whether the signals approach aligns with how people think about
> podling
> > > health
> > > - any obvious gaps or misinterpretations in the heuristics
> > >
> > > Note that it can also produce many dashboard-style displays and graphs
> of
> > > the data it has access to.
> > >
> > > Thanks,
> > > Justin
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: [email protected]
> > > For additional commands, e-mail: [email protected]
> > >
> > >
> >
>