@bryancsk 3. when you add your eggs back into the tomatoes, i do it at the VERY end so i can control their runniness. i'll pour on top, and then just let them steam ever so slightly (so they're still runny).
good luck 🫡
@bryancsk 1. egg seems overcooked. i'd cook the eggs in a seperate pan, 50%.
2. tomatoes seem undercooked? (i tend to let them cook through a bit more, and this is why i cook them separately). i want them to be almost falling apart (i usually use xiaoxing wine to help bring out more water)
@jiratickets one local one that people haven't mentioned, elkhorn slough. you can kayak out and literally see rafts of otters in the slough (and tons of other animals). highly reccomend.
@justalexoki okay but point 1:
He has shown you, O mortal, what is good.
And what does the Lord require of you?
To act justly and to love mercy
and to walk humbly with your God.
@TheGregYang@sergeantsup might be VOCs? a lot of furniture offgasses VOCs, and i've definitely had a headache from high levels (different people respond differently).
@burkeholland i've talked to so many customers who wanted to use jira's MCP server, and github's MCP server, and their lives all immediately improved when i showed them how to install gh/acli and how to write skills for them
@karpathy@airesearch12 being able to declare intent with specs was great, but we realized we wanted to do so recursively via a graph
we built https://t.co/6beVQFCRMZ to do exactly that! we wanted to capture intent (specs) and implementation details (issues) as a graph that's accessible to our agents
we saw something similar, where agents were way too fragmented to work on larger tasks (even when using planning documents)
we built https://t.co/VSsG2jOGms to try and alleviate this. instead of trying to guid an agent by conversing through codestacks, users can co-write specifications to recursively break down problems and hand them off to their agents.
@Andercot if you have well thought out ideas as the scaffolding, AI can actually write convincing ideas.
the hard part is having coherent thought out ideas though. most people fail here
@itsandrewgao we saw this happen too so we built https://t.co/6beVQFCRMZ to try and mitigate code-rot by treating specs as a graph that you commit to version control.
we wanted to provide traceability for what prompts generated what code, and a tool to use those specs as the driving layer.
@thdxr we designed https://t.co/6beVQFCRMZ to try and mitigate the think-rot claude code was giving us
ralph and beads only track surface level thoughts as tasks, but adding specs let us leverage deep thinking as a control layer for our agents. feel free to comment or try it out
@GergelyOrosz i've literally had this same exact thing happen to me, while making sales calls to companies that require zoom for recordings. no recording lol i just spam the transcript button as that's the only guarantee