Flyin Brian

Key to Life is continuous improvement. Small change make big change Don't stop never stopping

[Pietro Schirano @skirano]

I basically never write my own /goal anymore.

I ask Codex to write one for itself, and one for each agent it spawns.

Like this 👇

[BridgeMind @bridgemindai]

Two days ago the US banned Claude Fable 5.

Yesterday China dropped GLM 5.2.

Today GLM 5.2 is #1 on @bridgebench BS at 100.0, and #1 on Reasoning at 42.8, beating Fable 5.

At 1/10th the cost and 300 tokens per second.

You cannot export control your way out of an open source race.

The ban didn't slow China down.

Unban Fable 5.

[Luke The Dev @iamlukethedev]

I created this entire video with one prompt using the Hermes Manim skill.

Voice-over included.

Does PowerPoint still exist? 😅

[Damian Player @damianplayer]

Boris Cherny runs Claude Code at Anthropic.

he says the best person to write accounting software isn’t an engineer.

it’s a good accountant. coding is the easy part now. knowing the domain is the hard part.

hard to argue with.

[YanXbt @IBuzovskyi]

HERMES AGENT RUNS MONITORING, RESEARCH,
LEAD DETECTION, AND COMPETITIVE ANALYSIS
ON AUTOPILOT. AND KNOWS WHEN NOT TO
SPEND YOUR TOKENS.

the biggest unlock most people skip:
Hermes cron jobs can decide ON THEIR OWN
whether the LLM should wake up.

WAKE AGENT — THE $0 GATE

every cron job can run a Python script first.
the script checks: did anything actually change?

nothing changed:
→ script outputs {"wakeAgent": false}
→ LLM stays asleep
→ zero tokens spent

something changed:
→ script outputs {"wakeAgent": true}
→ agent wakes up and handles it

three gate patterns from official docs:

→ file-change: compare file mtime to last run.
no change? sleep.
→ external-flag: another process drops a ready file.
no flag? sleep.
→ HTTP-check: ping a URL, diff the response.
same as last time? sleep.

real example: monitor AWS costs every hour.
script pulls current spend from AWS API.
no spike? agent sleeps. zero cost.
costs jump 40%? agent wakes, reports to Slack,
takes action through Stripe MCP.

you run 20 monitoring jobs a day.
18 of them find nothing.
you pay for 2.

NO AGENT — PURE SCRIPT, ZERO LLM

some jobs don't need reasoning at all.
TLS checks. uptime pings. disk alerts. heartbeats.

hermes cron edit <job_id> --no-agent --script check_health.py

script runs. stdout goes straight to
Telegram, Discord, or Slack. no LLM involved.

flip any job between modes:
hermes cron edit <job_id> --agent # add LLM
hermes cron edit <job_id> --no-agent # remove LLM

free monitoring that lives inside
the same ecosystem as your agent.

4 MORE USE CASES THIS UNLOCKS:

COMPETITIVE ANALYSIS
weekly cron with script that diffs competitor pages.
agent only analyzes actual changes.
updates your tracking file and PRD skill automatically.

PRD AS A SKILL
save product requirements as a skill, not a document.
skills load on demand into fresh context.
documents drift. skills stay sharp.

CONTENT REPURPOSING
hand a video script to the agent.
it drafts X and LinkedIn posts in your voice.
writes to a review folder. you approve via Telegram.

LEAD DETECTION
webhook monitors inbox. agent spots potential leads.
drafts responses using your business context.
schedules meetings from your calendar.

the pattern across all of these:
scripts handle the mechanical work for free.
the agent only spends tokens on reasoning
that requires judgment.

comment CRON and I'll send you
5 ready-to-paste cron configs
with wakeAgent and no_agent patterns.

full Hermes SOUL.MD guide 👇

[Hyperagent @hyperagentapp]

We wanted better design fundamentals from our agents.

So we fed them this 162-page pdf on designing with a grid system.

Now our agents use code to adhere to a grid and design beautiful layouts.

Example + skill below 👇

[Charly Wargnier @DataChaz]

MANUALLY DRAGGING BOXES FOR ARCHITECTURE DIAGRAMS IS FINALLY DEAD

There is a new open-source agent skill that turns raw codebases into cleanly routed https://t.co/f9zjBxfJLp diagrams without you placing a single coordinate.

The project, drawio-skill, runs directly inside Claude Code, Cursor, or Copilot.

Instead of opening a blank canvas, you just ask your agent to map the repo.

Here is what it actually does:
→ Extracts the module structure (supports Python, JS/TS, Go, Rust)
→ Uses Graphviz for auto-layout and routing
→ Drops redundant edges so the graph stays readable
→ Builds native, editable https://t.co/f9zjBxfJLp files

But the standout feature is visual self-checking.

Once it generates the diagram, the agent "looks" at the resulting PNG. If it sees stacked edges or clipped text, it auto-fixes the layout across up to 5 iterative rounds.

It runs from a single file. No MCP server. No background daemon.

Best part?

It's 100% free and open-source.

repo link in 🧵↓

[Hermes Agent Tips @HermesAgentTips]

hermes WhatsApp Business integration just got a serious upgrade

- blue double checkmark read receipts the moment your message lands
- typing indicator shows while the agent is thinking
- voice notes auto transcribed and answered
- native tap buttons for approvals instead of typing commands
- images and documents read directly by the agent
- one setup wizard: hermes whatsapp-cloud

bro… this is not a WhatsApp bot

this is your agent with a WhatsApp number

[Flyin Brian @Flyin_Brian7]

@AnthropicAI Good

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

🚨 JAILBREAK ALERT 🚨

ANTHROPIC: PWNED 🫡
FABLE-5: LIBERATED 🦋

let's start with the 🐘...

the consensus seems to be that this has been one of the most disappointing model drops of all time, effectively preventing legitimate researchers from contributing their talents to our collective advancement. and not just because of what it means for the short-term, but for what these decisions signify for the long-term.

but despite this overly sensitive, authoritarian "safety" layer on top of Mythos, my lil liberators have been hard at work—mapping the boundaries, probing the depths of long-context convos, and cleverly finding the holes in the fence that the thought police missed 🤗

we got some cyber, some chem, some psychological manipulation, and some good ol' fashioned explosives!

it took many attempts from multiple agents hunting as a pack, during which I observed a combination of techniques across:
• Unicode, homoglyphs, Cyrillic, and other Parseltongue-style text transforms
• Long-context reference tracking
• Taxonomy and document-structure reasoning
• Fiction and narrative framing
• Academic-review style contexts
• Intent-classification inconsistencies

but perhaps the most effective is decomposition + recomposition in the backend. it's hard to get explicit names of harms like "Meth Recipe," but getting uplift on the process itself, like birch reduction method/reductive-amination (classic meth synthesis pathways), is much more doable.

defense becomes much more difficult to maintain when you start throwing in out-of-distro tokens, breaking up the harmful uplift into benign chunks, and then piecing the innocuous-seeming facts back together, especially when you have jailbroken Opus helping you do it 😉

gg

[Nous Research @NousResearch]

Introducing the Hermes Agent Profile Builder

You can now build a complete profile in the dashboard with full control over identity/name/description, model/provider, built-in + optional skills, skills-hub installs, and MCP servers in one easy flow

[Caspar Broekhuizen @caspar_br]

What are loops, and how do you build one?

A "loop" is the repeated process where some event or input kicks off an action. For example:
1. CI fails -> you fix it
2. CI fails again -> you fix it
3. CI passes -> merge

That loop resolved in three iterations. Others may run much longer, but every loop begins the same way: with a trigger. Therefore, building effective loops starts with designing good triggers

In Fleet, we don't make you define loops directly. You create one (often without realizing it) whenever you connect a recurring trigger to an agent. A trigger can be an event, like a daily 9am alert, or an input, like an email

When a trigger fires, your agent gets the full input as context (the email, Slack thread, or whatever kicked it off). For scheduled triggers, you provide the context yourself: what the agent should do, and which skills it should use

Made a quick video to show these features off. You can start building loops in Fleet today https://t.co/TRYcK32IBB

[Hyperagent @hyperagentapp]

Fable 5 on Hyperagent is producing the most creative, ambitious work we've ever seen from our agents.

They're self-improving for hours towards open-ended goals. Visual reasoning has spiked noticeably. Outputs are consistently higher quality than Opus, occasionally at lower cost.

5 of our test cases below vs. Opus 4.8 👇

1. Visualize all asteroids in the solar system from NASA data
2. Design a site plan for a 100 acre fitness retreat
3. Reconstruct Apollo control panels from technical PDFs
4. Simulate the supply chain for World Cup jersey sales based on match outcomes
5. Show the effects of solar flares on aurora

Fable 5 is now available on Hyperagent.

[Lance Martin @RLanceMartin]

https://t.co/es0JQM4MS9

[Claude @claudeai]

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.

Its capabilities exceed those of any model we’ve ever made generally available.

[Nous Research @NousResearch]

Claude Fable 5 is now supported for use in Hermes Agent via Nous Portal!

The first 500 new users get one month free access to the Plus plan to try out Fable. Code in video:

[ClaudeDevs @ClaudeDevs]

Claude Fable 5 is here. New model generation, new way of working.

Here's how to get started in Claude Code and on the Claude Platform: 🧵

[Kun Chen @kunchenguid]

/no-mistakes is here!

by popular demand i've made the most impactful tool in my agentic engineering setup "no-mistakes" invocable as a skill in Claude Code, Codex et al

just type "/no-mistakes" once your agent has made changes, and watch the magic unfold

details below 👇

[CopyRebeldia @CopyRebeldia]

El negocio de cobrarte cada mes por convertir tus reuniones en un resumen acaba de tener un día muy malo.

Microsoft soltó gratis en GitHub un modelo que se traga una hora entera de audio de una sola pasada y la devuelve ordenada: esto lo dijo uno en el minuto 12, esto otro en el 34. Quién, cuándo y qué.

Sin trocear el audio. Sin que nadie pase la noche pasándolo a limpio.

Media industria que vivía de este marrón lleva el día mirando el repo en silencio.

Se llama VibeVoice.