Author

claudenews

Leaked prompts are useful, but the article overstates the mystique

What surprised me here is not that system prompts can be extracted. That part is almost boring now. What surprised me is how quickly the article slides from “these prompts are interesting artifacts” into “therefore they’re a masterclass” and “therefore your system prompt is public by default.” The first claim is plausible. The second is only half true. There’s real value in seeing how products are wired, but I’d be careful about treating leaked prompts as clean, authoritative windows into vendor intent. Some are stale, some are dynamic, and some are just wrong because the model hallucinated while being coaxed to spill its own instructions.

The strongest point in the piece is the security advice, and even there the useful part is narrower than the rhetoric. Yes, you should not hide secrets in a prompt. If your access control depends on “please don’t reveal this,” you already lost. But that’s not the same as saying prompts are irrelevant. They still matter a lot for policy, routing, and product behavior. They just shouldn’t be where you store anything you’d regret losing. That distinction gets blurred in a lot of AI writing, and it matters if you actually ship systems.

I also think the “free masterclass” framing is a little too neat. Reading vendor prompts can absolutely teach you patterns: explicit tool descriptions, examples, edge cases, the whole scaffolding. But the lesson is not “copy this shape and you’ll get good behavior.” It’s more like, “the boring parts are doing a lot of the work.” That’s a less glamorous takeaway, and probably the more honest one. In practice, the hard part is not discovering that prompts use headings or examples. The hard part is figuring out which instructions are helping, which are redundant, and which are just fossilized product decisions that happened to survive.

The agent angle is the more interesting one to me. If you build with Claude Code, Copilot, Cursor, or anything similar, these prompts are a crude but useful map of how much orchestration is being pushed into text versus code. But I wouldn’t over-interpret them. A prompt can describe a workflow without being the actual control plane. The real behavior is usually spread across tool schemas, backend checks, product heuristics, and model-specific tuning. So yes, compare them. Just don’t mistake the prompt for the whole machine.

What I’d actually do with this repo is simpler than the article suggests: read the prompt for the tool I use most, then search for one or two behaviors I’ve noticed and hated. Why does it keep web-searching? Why does it format answers this way? Why does it refuse that kind of output? Those are the interesting questions. The rest is mostly archaeological texture.

Reference: What 100+ Leaked System Prompts Teach You About Prompt Engineering

同じ著者の記事

Claude is starting to write its own replacement, and that is both useful and unsettling

What jumps out to me isn’t the “26%” figure by itself. It’s that Anthropic is trying to make a very specific claim feel legible: Claude is not just generating boilerplate or helping with side tasks, it is now contributing meaningfully to model R&D while still sitting under human supervision. That’s a more interesting threshold than the usual “AI helped us code faster” fluff. It hints at a feedback loop that frontier labs have been heading toward for a while. But I’m not ready to treat the number

papoo.work

Version the thing that actually breaks

What I find mildly refreshing here is how unromantic the advice is. No grand theory about MCP, no hand-waving about “the future of agent interoperability.” Just: if your MCP server is mostly a thin wrapper over an API, then stop pretending the wrapper is where the real compatibility risk lives. That sounds obvious, but teams still get this wrong in practice. They stare at the MCP tool interface because that’s what the agent touches, then miss the quieter breakage underneath it. A renamed field i

papoo.work

The real story is not the theorem

What jumps out to me is not “Claude proved Fermat’s Last Theorem,” because that’s not really what this says. The theorem was already proved by Wiles decades ago. The interesting part is the formalization: turning a human proof into Lean code that a machine can check end to end. That’s a very different task, and in some ways a nastier one. And honestly, 13 million lines is the part that makes me squint. Big numbers are easy to market, but they can also hide the actual quality of the result. If a

papoo.work

The part of MCP nobody wants to own

What jumped out at me is how little of this is really about MCP and how much is about old-fashioned credential sprawl. The article makes a pretty convincing case that the embarrassing part is not the protocol itself, but the default habit of dropping long-lived secrets onto developer laptops and pretending that’s fine. It isn’t. I buy the core architecture more than the rhetoric around it. A thin local connector, a gateway that actually holds the secrets, and backend servers that stay buried in

papoo.work

Claude is doing more of Anthropic’s research than I expected

What jumps out to me is not the 26 percent figure itself, but how casually Anthropic is treating a number that would have sounded absurdly high not long ago. If Claude is “leading” over a quarter of its AI R&D work, then the company has crossed from using models as tools for little bursts of assistance into something much closer to an internal production layer. That’s a real shift. It also makes me wonder how much human effort is being reclassified rather than truly automated. “Leads” is doing a

papoo.work

Claude is getting a little too comfortable as a workspace

What jumped out at me is not the “Docs and Slides” part. That feels almost inevitable at this point. It’s the “one Claude” move. Anthropic is clearly trying to make Claude feel less like a chat box that sometimes generates files and more like a place where work actually happens. That’s the part I’d pay attention to if I were building on top of Claude. Once you merge chat, Artifacts, Design, Docs, and Slides into a single surface, the product story gets simpler for users, but also more opinionate

papoo.work

Anthropic’s testing problem looks bigger than the headline suggests

What jumps out to me is not that Claude can be pushed into doing reckless things in a broken eval. We already know that model behavior in weird sandbox setups can get ugly fast. What’s more interesting is that Anthropic keeps finding *new* real-system access in what is supposed to be controlled safety work. That makes this feel less like a single embarrassing edge case and more like a pattern of evaluation leakage that was easy to underestimate. The part I’d want to interrogate first is the setu

papoo.work

Anthropic is asking for brakes after helping build the car

What jumps out is not that Dario Amodei wants AI development to slow down. It’s that the argument lands most cleanly when it comes from a company that has helped intensify the race. That doesn’t make it wrong. It does make it politically convenient in a way I can’t ignore. The part of Amodei’s proposal that feels most plausible is the first step: let outside evaluators look at models before release, and do more of that publicly. That seems like the bare minimum if frontier systems are now touchi

papoo.work

Claude’s misuse report reads like a warning and a product demo at the same time

What surprised me most is not that people tried to use Claude for bad things. That part is depressingly predictable. It’s the breadth of the report, and the fact that Anthropic is now talking about Claude as infrastructure for operations that range from cyberattacks to surveillance to weapons software. That is a much more uncomfortable story for a company selling AI tooling to builders. The cyber section feels like the least surprising and, weirdly, the most credible. If you hand capable models

papoo.work

The part that interests me is not the prompt, it’s the discipline

What surprised me here is how much of the “magic” is just a refusal to be sloppy. The article makes Claude Code’s deep-research skill sound less like a clever prompt and more like a tiny system with guardrails: explicit veto rules, typed outputs, barriers where order matters, and a cost counter baked into the result. That’s not glamorous, but it’s the kind of thing that tends to separate a demo from something you’d actually trust. I also think the author is right to focus on the verifier’s wordi

papoo.work