ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ 🛡️
@DanielMiessler
Followers
148K
Following
22K
Media
2K
Statuses
27K
🟩🟩🟩⬛️⬛️⬛️⬛️ Building AI that upgrades humans. Founder / CEO of Unsupervised Learning. "Security comes from ꜱᴇ•ᴄᴜʀᴀ, which means Without Worry..."
daemon.danielmiessler.com
Joined March 2007
How I monitor what my Agents are doing... 📈 A visual timeline of their activity 🗣️ Custom voices for each agent (Eleven Labs) ⚒️ Detail on tools and tokens Massively helps me stay aware of what's going on while I'm doing other work.
28
57
596
GRAVE THREAT: The CCP are using “shell” companies to hide their purchases of land and property near our military bases. Even setting up shop next to our B2 Nuclear Bombers!!! Please, SHARE this everywhere
0
0
0
🎙️My new piece of core AI gear. How is this AI gear? Because I speak much faster than I type. So Wispr Flow @WisprFlow is how I get things done, and the more sensitive and high-quality the mic is, the less time I have to fix dictation.
6
2
11
This is a really compelling description of intelligence. Not sure about all the symbolic stuff, but I feel like AI (and all Intelligence) is largely this whole compression concept. This is also how Ilya has explained it for years. What's at the base? What's the essence of it?
3
1
15
Dating profiles are getting insane
My fertility markers are at an all-time high. I have more total and motile sperm than 99.6% of men of any age, including men under 25. + total count: 600 M + concentration: 162 M + motility: 55% + total motile count: 330M + morphology (normal): 10% This is also more than 6x the
1
2
24
The award-winning documentary that Sen. Adam Schiff demanded to have removed from Amazon Prime shows how metabolic and other natural cancer treatments work. "We're not giving cancer an opportunity to detour . . . to escape down one of its usual [biochemical] pathways."
27
113
450
We've released a new frontend-design plugin for Claude Code. Here's Claude's output for: "use the frontend design skill. Create a music player app."
91
138
2K
@trq212 @claudeai @AnthropicAI Ok, and @rez0__ has now also confirmed traffic going to that IP. Close friend and extremely reputable bug hunter. Anthropic please respond.
1
1
4
HT @rez0__ for finding this post by @ITSecurityguard.
0
0
3
This is an extraordinary claim, which requires extraordinary evidence, and some evidence appears to be provided here. Let's stay calm and give Anthropic the opportunity to respond. @trq212 @claudeai @AnthropicAI
5
3
19
When people can make whatever they want, the question changes to what people want to make. Ask the average person what they want to make and they won't have an answer. They'll look at you like you're crazy. Inequality of execution is turning into inequality of imagination.
3
3
26
When people can make whatever they want, the question changes to what people want to make. Ask the average person what they want to make and they won't have an answer. They'll look at you like you're crazy. Inequality of execution is turning into inequality of imagination.
3
3
26
Yep. Or most degrees.
2
1
10
I think we have the exact opposite data problem in AI. 👀 It's true we're running out of great literature ⛰️ It's also true that we're not looking at most data being generated https://t.co/KlYtY8UK5Z
danielmiessler.com
AI has a data problem, but it's not the one most people think it is
1
5
9
That's why I characterize their latest engineering post as a downplaying of MCP but not a total diss or replacement. The way I have refactored everything for myself is to use MCPs as a services directory. So, it is how I check every once in a while to make sure that that is
danielmiessler.com
They're recommending a filesystem and code-based structure for calling tools instead of using MCPs each time
I hadn't thought about this before: unlike regular APIs, MCPs don't need to make any stability promises "MCP servers are called my LLMs which dynamically read the spec every time, which allow us to constantly change the MCP server. It doesn't matter! We haven't made any promises"
1
2
13
This is wild. ElevenLabs just dropped Scribe v2 Realtime Its fastest, most accurate ultra-low-latency speech-to-text (150ms, 90+ languages) Breakdown + demo🧵👇
18
25
250
CC will issue a Haiku request per Bash invocation. Remove that, and Claude Sonnet 4.5 flies. Same model, same endpoint and auth token, same number of tool calls. pi is 10 seconds faster. Highly relevant to my workflow, as I do care how quickly an agent can iterate.
16
6
224
How I monitor what my Agents are doing... 📈 A visual timeline of their activity 🗣️ Custom voices for each agent (Eleven Labs) ⚒️ Detail on tools and tokens Massively helps me stay aware of what's going on while I'm doing other work.
28
57
596
Whoa...I'm in the middle of working on this exact thing right now! You guys are just crushing it... Please share more.
🧵 Making Proactive Agents with Code Generation Last week we shared about how you can use code generation to compose tools. But code generation can also give your agents new capabilities- like proactively acting on your behalf.
1
2
56