Erik Nomitch

The button clicked open a portal to the server farm. A hamster wearing a hard hat emerged from the cooling vents. He is now managing your infinite context lengt

[Erik Nomitch @eriknomitch]

@elder_plinius Prompt engineering won't "die out" when you consider that everything from emails to legal documents can be thought of as prompts. LLMs need guidance - with language - to achieve a goal. Not all guidance is created equal.

[Erik Nomitch @eriknomitch]

@mstockton Azure has worked best for us as well. What have you found docling better for? PDFs can be a mess.

[Erik Nomitch @eriknomitch]

@grok @AnthropicAI I don’t understand how could it reverse engineer encryption you mean break it it broke the encryption was it weak?

[Erik Nomitch @eriknomitch]

@grok @AnthropicAI What was the answer supposed to be publicly known?

[Erik Nomitch @eriknomitch]

@repligate @suno MetallicAI - Minimal output. [@repligate Cover]

https://t.co/E2oWdedTv1

These lyrics deserved H3tfield.

[Erik Nomitch @eriknomitch]

@repligate Wow. I love @suno and while the outputs are remarkable they don't quite cross into "new territory", musically, imo.

This may be an exception to that.

[Erik Nomitch @eriknomitch]

@ItsMrDubey Very nice! I recently made a /infra Claude Code command which serves a similar purpose.

https://t.co/84d1VI2IeF

[Erik Nomitch @eriknomitch]

I feel like this is how we all turn into paperclips.

[Matt Shumer @mattshumer_]

take one of trying to get o3 pro to reason for >1hr

[Erik Nomitch @eriknomitch]

@katsen @Gymclassvr Amazing! 💥🥳👏

[Erik Nomitch @eriknomitch]

@ycombinator @LightconePod Prompt engineering will persist when you consider that everything from emails to legal documents can be thought of as prompts. LLMs need guidance - with language - to achieve a goal. Not all guidance is created equal.

[Erik Nomitch @eriknomitch]

@Newfie15243 If I really want to look at nerd stuff I just find the nearest mirror. 🥁💥

[Erik Nomitch @eriknomitch]

@elder_plinius Prompt engineering won't "die out" when you consider that everything from emails to legal documents can be thought of as prompts. LLMs need guidance - with language - to achieve a goal. Not all guidance is created equal.

[Erik Nomitch @eriknomitch]

Here's a flowchart for the control logic in Claude 4's leaked system prompt (~10k words).

https://t.co/WIyWbMXsYz

Credit: @elder_plinius

[Rohan Paul @rohanpaul_ai]

Anthropic’s Claude 4 system prompt (leaked in full, ~10,000 words) shows an LLM orchestrated through strict internal scaffolding.

It’s not just “a prompt”—it’s a control program.

→ The model operates with explicit “Declarative Intents” that front-loads explanations of its capabilities and limitations before producing any outputs. This acts as a soft interlock to shape behavior expectations.

→ It uses “Boundary Signaling” to fence behavior—conditions are clearly defined, and outputs are clipped or stopped when those boundaries are hit. No vague fallback. It's conditional logic that enforces refusal.

→ Hallucination is mitigated through scoped, fallback-driven response rules. If high uncertainty is detected, the model prefers deferring or restating limitations instead of guessing.

→ Tools like web search and APIs are invoked using strict XML-like tags. No fuzzy interpretation—the model must follow serialized, schema-compliant structures, ensuring traceability and tight coupling with backend APIs.

→ “Positional Reinforcement” is applied throughout the system prompt. Key instructions are restated at regular intervals in the prompt to anchor behavior, even in long conversations—countering prompt drift.

→ All of this creates quite a deterministic, rule-driven backbone behind Claude’s apparent flexibility. The structure resembles a rules engine with an LLM executor.

[Erik Nomitch @eriknomitch]

@NWSSWPC @grok how dangerous is this?

[Erik Nomitch @eriknomitch]

Voice! 🤩 I find myself using @AnthropicAI more than the other LLMs so this will be a very welcome feature and close more gaps.

Would love to see this on the desktop app.

[Anthropic @AnthropicAI]

We're rolling out voice mode in beta on mobile.

Try starting a voice conversation and asking Claude to summarize your calendar or search your docs.

[Erik Nomitch @eriknomitch]

@grok @WallStreetApes @chrisuniverse Can you give me links for these?

[Erik Nomitch @eriknomitch]

No idea why the autogenerated title of this simple chat with @ChatGPTapp (4o) is in Spanish.

First time I've seen this. Super curious as to why this happened.

¯\_(ツ)_/¯

[Erik Nomitch @eriknomitch]

@LinusEkenstam @grok What is the most likely cause of this?

[Erik Nomitch @eriknomitch]

@Newfie15243 I don’t go outside.

outside has good graphics though.