Courtney | Opichi AI
Region: US
Tuesday 14 July 2026 19:18:16 GMT
6824
278
50
50
Music
Download
Comments
Brian 🇺🇸 :
This is why openspec is so useful.
2026-07-15 05:12:02
0
user8656871234304 :
But but if the.MD file had no teeth was just a suggestion why is the jason file better?
2026-07-15 02:02:45
1
CJCee :
The agents long for Jira
2026-07-14 22:58:20
3
Just Adam :
fable does that for me by rule when starting a project.
2026-07-14 23:07:16
1
Matt :
A better idea, build a console app, that the agent asks for next task info. It can add, update, modify those tasks. The json is hidden behind it, you have a small skill defining the tools usage. It only reads the context it needs for the task instead of reading all the information into its context space from a document.
2026-07-14 23:28:59
0
Cody :
I have my main agents connected to a task board, each project spawns high detail tasks to be accomplished and each respective task has its updates, current state and such. so as I am watching it progress through a large change it's checking off 10 to 20 tasks with high detail accuracy. each task simply lands in the done bucket with an entire history of all changes captured. that combined with GIT and infrastructure documentation my agents can return from a compaction fully functional in my environment without missing a beat
2026-07-14 19:46:34
1
PairOfBoxers :
I have my AIs connected to Jira. It creates ticket when it find issues. If I find a bug, I describe it to AI and it makes the ticket with the prompt. I tag appropriately and have the AI loop through the tags.
2026-07-14 22:42:40
1
Order-By-Chaos ♾️ :
Doesnt orchestrating Subagents (each with their own job) fix this?
2026-07-14 22:49:59
0
ChrisCavallucci :
CC turned to GitHub Actions to run tests. Nice for traceability but now it’s eating my budget like crazy. 😳 Suggestions?
2026-07-14 21:15:10
1
Paul Vazquez :
Define long build.
2026-07-14 22:14:18
0
mo_jeepn :
Or you could use the GSD (get-shit- done)plug-in in Claude Coach CLI. it’s more project related like what you’re talking about and I’ve had many successes on large projects. It is great for Greenfield work.
2026-07-14 23:55:28
0
Jay Kelly :
This is really, really, really sound advice.
2026-07-14 23:11:40
3
Cory Bohn :
Do you have a prompt for this? So when you’re working with Claude code, you can have them set this up.
2026-07-15 00:10:58
0
wesjofficial :
So where does the json worklist get placed in the project folder structure in the root? 😳
2026-07-14 23:02:50
1
🌶️ustin :
Can you implement this mid build or does it only make sense if you do it from the start?
2026-07-15 01:04:16
0
BCWerks :
Jira and GitHub Issues
2026-07-15 00:09:32
0
It’s Nicon :
No — keep markdown, unless a program actually parses the file. JSON only earns its cost when something machine-reads it and enforces state; otherwise it’s ceremony that costs you rationale, comments, and readability.
2026-07-15 00:07:59
0
Romero :
who is Jason?
2026-07-14 23:29:11
0
Kapil MM :
so true
2026-07-14 21:55:31
1
Nova3 :
This is still the operator doing overwatch on itself, you need to move it out of the thread completely and do deterministic auditing. But remember, the minute the deterministic work re-enters the operator thread, the probability starts all over again. There’s no such thing as 1.0 in a probability model.
2026-07-14 23:25:04
0
Kevin Hoff :
New to this? You don't have to wire anything up yourself. Tell your coding AI: "add kenoodl Verify to my project" and it does the rest.
2026-07-15 04:40:50
0
To see more videos from user @opichillc, please go to the Tikwm
homepage.
