trees101

I'm interested in this, what is the best current method/tool for this task?

you can copy extensions rather than install them to avoid that problem

Web clipper is great. just tried the reader mode on this chatgpt transcript, it only shows 1 page of it. Is the purpose of reader mode to enable interactive annotation before saving to notes?

what is a good way to read PDFs using AI?

there will always be a difference between the general capabilities, and the particularities of your exact environment and requirements.

Closing this gap is done in the harness, either through Skills, user behaviour/prompts , Agents.md etc etc.

I think that this is an area worth investing time in, but it is indeed hard to know what the scope of this is.

can you please share details about your harness

Anyone use this stuff with Delphi? I've been looking for tips for getting the best out agents for Delphi

looks like you've done some thorough testing. Have you found that prompting reliably reduces premature quitting? And have you found that reducing premature quitting results in more accuracy?

From my reading, the official docs don’t support the strong claim that frontier LLMs are explicitly RL-trained to “be lazy” or conserve tokens as claimed in this thread. What they do document is adaptive / hidden reasoning compute: OpenAI says reasoning models allocate internal reasoning tokens and reasoning.effort controls how many are used (https://developers.openai.com/api/docs/guides/reasoning), and Anthropic says adaptive thinking decides whether/how much to use extended thinking based on request complexity, with effort as soft guidance and max_tokens as the hard cap (https://docs.anthropic.com/en/docs/build-with-claude/adaptiv... hinking). So prompt wording may change how the same budget is spent, but it can’t exceed the hard token cap.

Also, the “encouragement helps” anecdote seems real in the AlphaEvolve workflow, but I can't see that forpublic models. Gómez-Serrano says this in Quanta (https://www.quantamagazine.org/the-ai-revolution-in-math-has... rived-20260413/), and the released AlphaEvolve notebooks really do contain prompts like “Good luck, I believe in you...” (https://github.com/google-deepmind/alphaevolve_repository_of... oblems, e.g. https://github.com/google-deepmind/alphaevolve_repository_of... blems/blob/main/experiments/finite_field_kakeya_problem/finite_f ield_kakeya.ipynb). But those prompts also bundled strong structural hints (“find a general solution”, “better constructions are possible”), so from my reading the evidence is: prompt phrasing matters, especially in an internal search stack, but not “pep talks are a universal reasoning hack.”

The P≠NP conjecture in CS says checking a solution is easier than finding one. Verifying a Sudoku is fast; solving it from scratch is hard. But Brandolini's Law says the opposite: refuting bullshit costs way more than producing it.

Not actually contradictory. Verification is cheap when there's a spec to check against. 'Valid Sudoku?' is mechanical. But 'good paper?' has no spec. That's judgment, not verification.

Skill issue. I'm far more interactive when reading with LLMs. I try things out instead of passively reading. I fact check actively. I ask dumb questions that I'd be embarrassed to ask otherwise.

There's a famous satirical study that "proved" parachutes don't work by having people jump from grounded planes. This study proves AI rots your brain by measuring people using it the dumbest way possible.

oh I see, you're force-revoking someone else's key

why would you do that rather than just revoking the key directly in the anthropic console?

great tips if you want more context, aider has /copy-context that copies the files that you have added to the context, your chat (I think). you can then paste into a subscription chat app where you're not paying per token

https://github.com/hotovo/aider-desk is a gui, takes 5 mins to install, has MCP support (try context7). Definitely worth a look and is an "easy" way in to aider.

its a pity that it only works with Claude out of the box. There is a way to proxy it to other models: https://github.com/1rgs/claude-code-proxy I've found it works with Gemini. But would be better if it just allowed switching.

how do we access this?

I tried it and it didn't work too well. I suspect the prompts were optimized for Gemini, not local Gemma.

TBH I found the whole thing quite flaky even when using Gemini. I don't think I'll keep using it, although the concept was promising.

edit your `.config/tmuxai/config.yaml`

to add these lines:

``` openrouter: api_key: "dummy_key" model: gemma3:4b base_url: http://localhost:11434/v1 ```

Not sure how accurate my stats are. I used ollama with the --verbose flag. Using a 4090 and all default settings, I get 40TPS for Gemma 29B model

`ollama run gemma3:27b --verbose` gives me 42.5 TPS +-0.3TPS

`ollama run gemma3:27b-it-qat --verbose` gives me 41.5 TPS +-0.3TPS

Strange results; the full model gives me slightly more TPS.

very interesting, I'm looking for something like this, will check it out. In my obsdian vault, I keep a lot of code snippets and even entire python scripts. Do you see your method as being perhaps an alternative to dedicated github repos for tiny personal projects, replacing a million little repos? And at the same time having notes in the same repo?

Have you solved the git repo index problem, Ive found that large vaults cause a problem and require occasional cleanup:

```bash git gc --prune=now # Garbage collection

git repack -ad # Repacked objects, optimizing repository storage ```

with Claude coder, how does history work? I used it with my account, ran out of credit then switched to a work account but there was no chat history or other saved context of the work that had been done. I logged back in with my account to try copy it but it was gone.

with Claude coder, how does history work? I used it with my account, ran out of credit then switched to a work account but there was no chat history or other saved context of the work that had been done. I logged back in with my account to try copy it but it was gone.

Can you expand on that? Where do big enterprise orgs products fit in, eg Microsoft, Google? What are the leading providers as you see them? As an outsider it is bewildering. First I hear that llama_index is good, then I hear that its overcomplicating slop. What sources or resources are reliable on this? How can we develop anything that will still stand in 12 months time?

Does this require you to run a virtualized apple OS in order to keep track of your tags?

fair point

how is that better than AI Coding tools? They do more sophisticated things such as creating compressed representations of the code that fit better into the context window. E.g https://aider.chat/docs/repomap.html.

Also they can use multiple models for different tasks, Cursor does this, so can Aider: https://aider.chat/2024/09/26/architect.html

use https://github.com/Aider-AI/aider clone any repo, use aider for Q&A about the code. Use aider to add features to the repo and do experiments with the code. Its a very interactive way to learn

If you're searching across a bunch of open-source repos, Grep.app can be super handy—it’s fast, supports regex well, and lets you filter by language and license. But if you’re focused on a specific GitHub repo, GitHub’s native search is probably better. It ties in with commit history, issues, and PRs, and the new symbol search is great for navigating large codebases. Basically, Grep.app is good for broad searches across projects, while GitHub search is stronger for digging deep into one project.

looks like a nice tool for evaluating code blocks directly in Vim's Markdown buffers, and the ability to redirect outputs into named blocks is a cool way to keep everything contained in the editor. If you're looking for something similar but more versatile across different environments, check out *Cog*. It embeds executable code in any text file and inserts the output back into the document, which is great for automating documentation outside of Vim, especially in CI pipelines.