maikSan

[B 612 @Witch_B612]

Nunca foi tão fácil escolher
LULA13
#LulaVerdeAmarelo

[Bikash | Hermes Super-Intel @BkashJosi]

My Hermes folder was 33 GB. I built Footprint to find out why. 🪽

Agents leave behind more than chat history: toolchains, worktrees, backups, caches and the files they create.

Footprint puts that storage into a visual map inside Hermes Desktop, organized around what each folder does.

On my machine:
→ Python package cache: 1.03 GB, marked as rebuildable.
→ An old worktree: 2.26 GB. Its commits were on a remote, but 7 untracked files meant “look first.”
→ Bot activity: 407 recorded file writes, with a breakdown by bot and folder.

That worktree is why size alone isn’t enough to decide what to remove. Backed-up commits don’t account for untracked files.

You can inspect the cleanup candidates or use “Hand to a Bot” to draft a specific cleanup job in chat for approval.

My routine: inspect the folder, review what would be removed, then check the result after cleanup.

Inspired by @tobi’s disktree, built specifically for Hermes.

Free and open source. Repo and recording tool in the first reply.

What’s taking the most space in your Hermes folder?

[Gabriel Adamuchi - kairogen.ai @adamuchigabriel]

Um desenvolvedor liberou centenas de skills gratuitas para IA: 63 antes e 249 skills.

O nome é ECC e é gratuito.

O projeto é aberto no GitHub, funciona com Claude e ChatGPT e já passou de 240 mil estrelas.

Comenta "ECC" para receber o link.

[Bikash | Hermes Super-Intel @BkashJosi]

10 Hermes commands for the moments your agent gets stuck, drifts or needs a do-over.

One useful command to try:
/steer Focus on the login bug. Leave styling unchanged.

The official docs say /steer passes that note to the agent after its next tool call. Check the diff afterward: did the work stay within your scope? Steering is an instruction, not a file-access restriction.

Here are 10 documented commands to keep nearby:

→ Diagnose setup: hermes doctor, hermes logs
→ Inspect context size: hermes prompt-size, /context all
→ Manage a long chat: /compress, /branch
→ Back up Hermes or restore checkpointed files: hermes backup, /rollback
→ Guide this turn or the next: /steer, /queue

Use /queue for a separate instruction you want handled after the current turn.

Run “hermes …” in your terminal and slash commands in a supported Hermes chat. Check your version and client. /rollback needs checkpoints enabled before the change; coverage is limited.

Which should get a walkthrough: steering a code fix or restoring a checkpointed file?

[Chen @ch3nweiii]

this is pure f*cking treasure

a Stanford AI research group just figured out how to run Claude Opus 5.5 and GPT-6.1 Sol as one team, and it breaks straight through the search scaffolding wall

most developers combine frontier models in one of two ways:

-> chain prompts inside a single context window and burn tokens on bloat
-> hardcode an evolutionary loop and freeze the models into search rules that go stale within 5 iterations

Stanford deletes the human-designed scaffolding and splits the stack in two:

> search director: Claude Opus 5.5 plans the search on top of a persistent idea-graph, where MAP-Elites and MCTS collapse into single Cypher queries
> candidate proposer: GPT-6.1 Sol explores high-entropy code variations inside isolated sandboxes
> no context bleeding: every session resets after each iteration, the graph keeps the lineage
> evaluator isolation: candidate code never touches the scoring process, so reward hacking and test leakage die on arrival
> meta-agent distillation: an offline pass pulls out verified lessons and updates search guidance across every branch

and the numbers beat the standard discovery baselines:

> 3.2x lower model spend than fixed evolutionary frameworks
> 1st place across 7 competitive AtCoder heuristic contests against human competitors
> above published SOTA on Anthropic's kernel builder and 11 mathematical optimization tasks
> 0% reward hacks: candidate-controlled metric tampering is gone

Opus directs the graph. Sol explores the code.

stop writing rigid search harnesses. let the frontier models own the discovery loop.

[Bober_smart @Bober_smart]

It is becoming f*cking dangerous: I launched a full-fledged autonomous crawler that parses any website with ZERO LINES OF CODE

Forget about Python, terminals, CSS selectors, and constant manual copy-pasting

You can set up a seamless No-Code integration just once, where Claude becomes the brain of your automatic data collector, and the results automatically flow straight into your spreadsheets

In this setup, a No-Code service accesses the website and grabs the raw content, while Claude instantly cleans out the noise, extracts the key insights, and organizes everything neatly

How to set this up:

1. Pick a No-Code service:
Sign up for makecom, zapiercom, or install the bardeenai extension (all of them offer free tiers)
2. Set up the Launch:
Specify how often the crawler should run (for example, Every day at 9:00 AM or At the click of a button) and feed it the target URLs
3. Connect the Claude module:
Add an AI step to your scenario and give Claude a clear instruction:

"Analyze the provided page text. Extract product names, prices, discounts, and seller contact details. Return the data as a clean, structured list"

4. Choose where to save the results:
Connect the final step (for instance, Google Sheets or Notion)

Quick 10-minute setup, full 24/7 autonomy, and zero headache with code maintenance. Essentially, with just a few clicks you are creating your own smart AI agent that quietly handles the routine work while you focus on real tasks

Bookmark this for later

[Nous Research @NousResearch]

It's time to connect 🌐

[Teknium 🪽 @Teknium]

Hermes Agent v0.22.0 is now live!

The highlights:

A much smoother easier to use onboarding - you don't even need to hook up a provider to get started - just install and start working!

Connectors - much easier to connect with any app, just click install and login - and done.

Stable Channel for Hermes Update - no more update anxiety and get releases that are tested much more thoroughly than a random commit, or switch back to main channel if you want the bleeding edge.

Over 2,500 people contributed to this version, thank you all!

See the full changelog here, it's our biggest version over version change ever:

https://t.co/rOPwklptGI

[My Pool Painter @mypoolpainter]

Meet LunoMac 🚀

A simple macOS utility to keep your Mac clean and under control.

Find unnecessary files, remove unused apps, explore disk usage, and keep track of system performance.

Everything you need in one place.

Clean. Native. Lightweight.

[Tony Simons @tonysimons]

Oh, I'm DEFINITELY testing this one. 🪽

Grok Voice inside Hermes Desktop.

You talk to Grok.

Grok hands the real work to Hermes.

Hermes does it, and the results get spoken back to you.

@julientalbot974 already showed the full voice-to-agent-to-voice loop working.

Now I need to hear it running on my Mac mini. 🔥

[Klonzu @Klonzu]

holy sh*t. 10,607 https://t.co/hIFHdQiZqi launches in 24h, 7,130 blocked, 56 trade signals. Stop letting one agent pick the entry

the trick is the rule in the header. every launch ends one of three ways: +60% first, -30% first, or neither in 60 min

orb trace RUN-084199 --from-buy

the scout reads 44 events per second
the cartographer links wallets into 1,271 groups
the guard blocks 73% of what reaches it

the bot only needs a human at three critical moments:

→ at the guard: $BONBIT gets blocked because 45% of its buyers share one funder. who decided 45% is the line?
→ at the quant: net EV -14.6% after 3.9% costs, so it skips. is +3% after costs the right bar?
→ at the label: 1,345 outcomes measured today feed a model still marked "in training". when is the sample big enough?

seven signals. three outcomes. one decision record.

This is Jev engineering one layer down. Of the last 70 launches it measured, 34 hit +60% first, 23 hit -30% first, 13 did neither. The decision stack ends at 6 entries, 68 skips and 14 blocks, so almost all the work is saying no

label first, block second, enter last

- the full bot setup
> 6 agents: scout, cartographer, judge, quant, guard, executor
> 7 signals: launch rate, curve inflow, followed flow, bundled share, creator risk, graduation rate, market phase
> 3 outcomes per launch: +60% first, -30% first, neither in 60 min
> skip rule: EV below +3% after costs
> executor cap: 1.8% in play, 15% limit

paste this prompt into Claude Code below:

"Build a research-mode decision layer for new https://t.co/hIFHdQiZqi launches in my project. No real trades:
1. Label every launch with one outcome: +60% first, -30% first, or neither in 60 min
2. Split the work into six agents:
> scout, cartographer, judge, quant, guard, executor
> each one writes a line to a shared decision record
3. Block a launch when too many buyers share one funder, and log that share
4. Skip anything with expected value below +3% after costs and fees
5. Paper executor only, with a hard cap on how much is in play
Show the plan and every file diff first. Do not apply edits until confirmed"

↳ the dashboard is a concept render in research mode with sample numbers, not a live account

[Bikash | Hermes Super-Intel @BkashJosi]

Pick up yesterday’s Hermes project without hunting through mystery chats.

Start by giving the session a name you’ll recognize:
/title Login bug investigation

Then use /resume to find it when you’re ready to continue.

That’s the easiest entry point to this 25-command Hermes Desktop cheat sheet, grouped around five jobs:

→ Remembering useful instructions
→ Organizing chats
→ Managing context and summaries
→ Working through tasks
→ Reviewing and recovering

For a longer project:
• /branch opens a separate chat to explore another approach.
• /compress summarizes conversation context when it gets crowded.

The sixth card picks five commands to start with. Try one on a small session, then check what changed before adding more to your workflow.

Availability depends on your Desktop and backend builds. Use /help to check yours. /undo removes a chat exchange; it doesn’t revert file edits.

Which loses you more time: finding an old session or repeating its instructions?

[Hermes Agent Tips @HermesAgentTips]

heads up... Hermes v0.22.0 now auto-deletes ended sessions after 90 days of no activity

pinned and open sessions are safe

if you'd rather keep everything, set this in your config:

https://t.co/4fljk1oXcE_prune: false

are you keeping all your old sessions, letting them go, or pinning the important ones?

[JOVEM PELA 4ª VEZ 🇧🇷 @JovemPela4aVez]

BORA LULA
BORA VIRAR ESSA ELEIÇÃO

#LulaPeloFimDaEscala6X1
#NinguémCalaOExercitoDoLula

[Bikash | Hermes Super-Intel @BkashJosi]

Installing Hermes is one step. Getting it connected, running locally and updating reliably takes more work.🧚

Hermes Agent v0.22.0 tackles those gaps. The major changes:

1. Easier first launch – Desktop setup becomes a guided chat with a one-time free trial model. No API key needed to begin.

2. Local models in one click – Hermes manages llama.cpp, checks which models fit your hardware, and handles downloads.

3. App connectors – Connect apps such as Notion and Linear. Multiple accounts are supported.

4. More controlled updates – Source installs follow tested stable releases by default. Dependency environments are checked before switching.

5. 600 reviewed plugins – Browse the catalog from Desktop or CLI. Several memory integrations move into plugins.

6. Browser login and human takeover – Password-manager integration keeps passwords out of the model's context. Bot Screen lets you take over for login, 2FA or CAPTCHA.

7. Better agent visibility – Watch, steer and stop subagents live. Multiple profiles share one gateway and Desktop backend.

Before updating: automatic pruning now deletes ended sessions inactive for 90 days. Set `https://t.co/eZO711BS7K_prune: false` to keep them.

This release also includes the earlier v0.21.x changes.

[vgenv | Affordable AI Video Generation @cySG20]

@nima2qisi2 免部署这条思路把门槛真的打没了，一个图一个指令就出环绕，谁都能上手😂 想再拉一档质感的用 https://t.co/dyBCfH95x1 云端 H3 跑同一招，0.01/s 还有每天免费额度 https://t.co/DrYyAP1Je5

[Dmiao豆苗啊 @nima2qisi2]

😆现在制作 360° Orbit 环绕效果太简单了
上传你的图片→ 输入指令 →玩法Get

Pormpt指令👇

[Tech Logic @The_techAI]

50 legendary websites that feel like the internet’s hidden toolbox 🧰

1. https://t.co/cr8PmdXzQ3 — free research papers
2. https://t.co/u1Y4OQzofE — papers when unpaywall misses
3. https://t.co/AMiFzHhpIM — borrow books
4. https://t.co/hSlNyaVFy2 — 70k+ free books
5. https://t.co/TF13aVtirR — audit university courses free
6. https://t.co/Q2f6XIBALL — step-by-step math
7. https://t.co/BA6QostDUE — background removal
8. https://t.co/ydt51BpHqu — object removal
9. https://t.co/Vw3lkNr98m — quick file tools
10. https://t.co/wM6uNsrmeS — PDF merge/split
11. https://t.co/Jpupdx9qmB — free Photoshop alternative
12. https://t.co/ekQAaVrz4o — image compression
13. https://t.co/9Sct0vy4xG — fast diagrams
14. https://t.co/Z3clM9gj1I — beautiful code screenshots
15. https://t.co/ZHNm8UtTwg — test regex
16. https://t.co/NmJPmKVLbS — clean JSON
17. https://t.co/TXTGrbnSVV — terminal commands explained
18. https://t.co/Fp4DfBwP53 — software alternatives
19. https://t.co/GFZpP7IK0s — breach check
20. https://t.co/ttAdldHv9q — temp email
21. https://t.co/o8EpsLG8LA — self-destruct notes
22. https://t.co/odr3ZJ78lF — live global radio
23. https://t.co/Co1DivGStz — focus sound
24. https://t.co/v6BDQ5YkCW — custom ambient
25. https://t.co/ghTsTt0LM5 — where to stream
26. https://t.co/7o82rjIOvw — digital library
27. https://t.co/65Zriz1VQ9 — old web pages
28. https://t.co/d7bpsRDuZE — site traffic
29. downtimerobot — uptime check
30. https://t.co/bJOSaEgehr — speed test
31. https://t.co/YqJNi8gdJC — compress images
32. https://t.co/hin3e40dIA — file conversion
33. https://t.co/AmhFummB77 — compare text
34. https://t.co/zT8moYe9Tg — disposable inbox
35. https://t.co/NHxVRFXhpQ — JS obfuscate
36. https://t.co/Oy5YoomuVb — typing practice
37. https://t.co/Sj8KlIeTnT — typing test
38. https://t.co/uRa33xAC0I — spaced repetition
39. https://t.co/GArJhJF9Vv — all-in-one notes
40. obsidian.md — local knowledge base
41. https://t.co/2O3e3sNetO — bookmark manager
42. https://t.co/f4zSErNFBF — task list
43. https://t.co/viUEdYgWad — boards
44. https://t.co/3ffAd1B7wU — whiteboard
45. https://t.co/Eck5egOndx — design
46. https://t.co/PQIRy3PfB4 — quick graphics
47. https://t.co/BA6QostDUE — again because it's that good
48. yt-dlp — video download (open source)
49. https://t.co/UXBTNgdpnw — video compress
50. https://t.co/evkkoH9QnC — local file transfer

Save this. You’ll definitely need some of these later. 🔖
Follow @The_techAI for more useful websites, AI tools & tech resources.

[Andrew Bolis @AndrewBolis]

Most people treat Perplexity like Google.

And they skip its powerful research features.

Perplexity is a full research engine built for depth and accuracy.

To use it effectively, start with the right Model:

Think of these as different "brains," each specialized for a specific type of work.

1. Sonar
Fast, source-backed live search.
Use it for real-time facts and breaking updates.

2. Claude Sonnet
Balanced, structured reasoning.
Strong in writing and summarizing.

3. Claude Sonnet Thinking
Slower, deeper step-by-step logic.
Best for detailed analysis.

4. Gemini Pro
Handles text, images, data, and math.
Great for technical or visual tasks.

5. GPT
Versatile and quick for general tasks.
Works well for everyday workflows.

6. GPT Thinking
Designed for long, structured reasoning.
Ideal for strategy and complex planning.

7. Grok
Casual, fast, and culture-aware.
Good for quick explanations and lighter queries.

Then use the right Features:

1. Pro Search
Instant answers with citations and live results.
Prompt: "Summarize recent changes to US tax laws with citations."

2. Research Mode
Builds organized, multi-step research automatically.
Prompt: "Write a comprehensive market analysis on the rise of vertical farming."

3. Pages
Turns chats into clean, formatted reports.
Prompt: “Turn this discussion into a blog post about productivity hacks."

4. Spaces
Shared folders for saved threads and notes.
Prompt: "Create a Space called 'Q4 Marketing' to track our campaign research."

5. Internal Knowledge Search
Search your own documents + the web together.
Prompt: "Using our brand guidelines and web trends, draft a social media strategy."

6. File Uploads
Upload PDFs, slides, or videos and extract insights.
Prompt: "Analyze these three earnings call transcripts and compare the revenue growth."

7. Labs
Quickly turn data into small dashboards.
Prompt: "Use Labs to turn this sales data into an interactive bar chart."

8. Tasks
Automate recurring updates in your inbox.
Prompt: "Every Friday at 9 AM, email me a summary of major crypto price moves."

9. Focus Search
Filter results by type or source.
Prompt: "Search academic papers for the latest studies on intermittent fasting."

10. Personalization & Memory
Recalls your role and preferences.
Prompt: "Set my default location to 'New York' for all local market searches."

11. Watchlists
Track topics, companies, or trends.
Prompt: "Add the NY Knicks and LA Lakers to my Sports watchlist."

12. Connectors
Pull data from Gmail, Drive, or Calendar.
Prompt: "Find the flight confirmation PDF in my Gmail from last month."

13. Assistant
Writes emails, plans meetings, and handles small tasks.
Prompt: "Write a polite decline for the invitation and suggest a reschedule for next week."

Now you can use Perplexity’s powerful research engine.

📌 Learn 30 free AI tools in 30 days: https://t.co/7MJCwAPtaQ

👉 Follow me @AndrewBolis for more and 🔄 Repost this to help others use AI

[delost @thedelost]

THIS IS ACTUAL F*CKING GOLD

Anthropic just shared one of the best tricks for Claude Code.

The idea: run Sonnet, Haiku and Opus 5.5 together, each one doing only the job it's best at.

Opus-level judgment on every big call, without paying Opus rates for every routine turn.

You just launch Claude Code like this:

claude --model sonnet --advisor opus

here's the split:

→ Sonnet 5.5 runs the session and writes the code
→ Haiku 5.5 explores files and docs in parallel
→ Opus 5.5 stays on call as advisor for the big calls
→ you stop spending Opus on work that doesn't need it

Opus only steps in at three moments:

→ before a plan locks
→ when the same error shows up twice
→ before calling it done

Sonnet builds. Haiku explores. Opus reviews.

JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split

the full tree
> Sonnet 5.5 runs the main session and edits code
> search maps the repo on Haiku 5.5
> grep finds call sites on Haiku 5.5
> docs pulls the API reference on Haiku 5.5
> Opus 5.5 on call as advisor, reads the whole transcript when consulted
> JEV routes micro-forks in milliseconds

paste the tree and this prompt into Claude Code ↓

"Rebuild my Claude Code setup around this tree:
1. Run the main session on Sonnet 5.5 with Opus 5.5 as advisor.
2. Create three read-only Haiku 5.5 subagents: search (maps the repo), grep (finds call sites), docs (pulls the API reference).
3. Add one rule to CLAUDE.md: consult the advisor before locking a plan, after the same error twice, and before calling a task done.
4. Route fast deterministic forks (tool choice, path selection, retries) through JEV.
Show me every change as a diff first. No edits until I say go."

↳ https://t.co/xb2zLxajd4

save it, you'll use this every single day.