Hey guys, need some help. I am doing too much with this: 1) apply various source/evidence heirachies tags to various research inputs 2) figure out the best way to incorporate into research (.md) outputs for obsidian, perhaps using some type of metadata or tags in line with text...I still havent figured out the best solution 3) include it as an operating contact for my research pipeline without boggling down the operations and context There is lots of concepts i created with claude here, and worried its now too "overburdening" What types of contracts do you guys have for research type pipelines, where you want to document evidence, score its source, and assert claims from truths etc. Can someone give advise on how to slim it down to the minimal required operating model for this purpose? *See my current file attached: since I clearly had claude write this up, its a pain to read and review...feel free to parse it through your own LLM, and give suggestions/revision otherwise. Thanks for your time Fam!

Pre-face: include lots of venting, do not continue further if you don't want to hear. Topic: AI producing too much text that is difficult to absorb. The more issues/iterations I raise, the more complicated and over-egnieered the solutions. I force myself to use AI tools minimum 4-5 hours a day, its especially hard at my stage in learning. Mental health is in decline. Diagnosis: I am a non-technical user, which is likelky contributing to much of my pain using AI. AI responses, plans, reviewing plans with so many complex ideas and procedures.....following into the trap of allowing the AI to build while i miss key assumptions. Impact: I am tired, constantly researching and asking for clarification or simplication. Or review this plan, make it better loop....I don't want to say Im going to wait for Mythos 9.7 before I can be effective. I know other non-technical users are suceeding and I need tips to join you. How to help non-technicals: I assume many non-technical users on here same experience as me. If i was native to software and technology terms, maybe its easier to skim and direct the AI. Mostly, the AI is not getting much direction because I am so consumed by endless comprehension exercises. Funnily enough, when I super focus and actually absorb it's output, I often realize it's just "strangely" worded. Not illogical, because the AI is very logical. Just not absorbable. I prompt "explain this simply to a 10 year old" like 50 times a day. No joke. Another one of my strategies is asking for Mermaid flow diagrams in obsidian, pre-facing every operating model file. This has helped a lot...any other tips? Who has actually experienced, but evolved past this stage? Lightbulb moment: I think I need to build a LLM WikkI with a pipeline that parses all my chat conversations into it....

The benchmarks are in, the community is super hyped. I've used probably 500~ prompts each on both new 5.6 and Fable. Honestly, I cannot with words describe the difference. I think its mainly because Im working with .md text files, not software coding. How would you describe the difference in words?

Hi Fam. Any Hermes users can assist with local connection issue? I am trying to stabilize a recurring Hermes Desktop connection issue that has been happening for months. Claude/Codex need it fix it on nearly every bootup :( The Windows Hermes Desktop app is configured in **remote mode**, not local gateway mode. It connects to a Hermes gateway running on a Hetzner VPS through a local SSH port forward. The model: Hermes Desktop on Windows > Windows SSH local port forward > SSH into Hetzner server Hermes > Hermes gateway on VPS lookback via SSH (I am non-technical, so if this is not correct way tell me!) When the issue happens, Hermes Desktop shows: > Hermes couldn't start > The background gateway didn't come up > Your remote gateway session has expired The immediate recovery is usually: 1. Restart the SSH tunnel. 2. Refresh the Electron remote OAuth/basic-auth session. 3. Relaunch Hermes Desktop. This works, but it keeps recurring. Claude's wording on what to ask my humanoid fellows: "My own strongest recommendation to ask humans to validate: use a supervised tunnel layer, probably `ssh.exe` wrapped by WinSW/NSSM with `ExitOnForwardFailure=yes`, `ServerAliveInterval`, and automatic restart, or replace the SSH tunnel with Tailscale/WireGuard so Desktop points at a stable private address. Ask the human experts in your community if this approach angle is correct." Thanks All.

This is probably well discussed given the rising popularity of ICM and Obsidian brains, like the one Jake recently discussed in this weeks High Tea. Do you guys have any prompts/skills to share to best dissect various types of media and parse it into your Obsidian library? I am currently not happy with the writing prose style, depth / how the information is structured as its being taken from input content. I am currently building the ICM input/output stages to adapt a parsing engine for finance/news/tech/software learning types of content into my "2nd brain". Perhaps I need to work on specific parsing logic for each. But its seemingly tedious and Im having trouble with getting a consistent output format that is satisfactory. Why? Because every case is unique...I mainly just want it to be as concise as possible, and have well structured themes.