The Claude Code Source Leak: fake tools, frustration regexes, undercover mode

For me one of the most interesting aspects is how compaction works. It turns out compaction still preserves the full original pre-compaction conversation in the session jsonl file, and those are marked as "not to be sent to the API". Which means, even after compaction, if you think something was lost, you can tell CC to "look in the session log files to find details about what we did with XYZ". I knew this before the leak since it can be seen from the session logs. Some more details:

  The full conversation is preserved in the JSONL file, and messages
  are filtered before being sent to the API.

  Key mechanisms:

  1. JSONL is append-only — old pre-compaction messages are never deleted. New messages (boundary
  marker, summary, attachments) are appended after compaction.
  2. Messages have flags controlling API visibility:
    - isCompactSummary: true — marks the AI-generated summary message
    - isVisibleInTranscriptOnly: true — prevents a message from being sent to the API
    - isMeta — another filter for non-API messages
    - getMessagesAfterCompactBoundary() returns only post-compaction messages for API calls
  3. After compaction, the API sees only:
    - The compact boundary marker
    - The summary message
    - Attachments (file refs, plan, skills)
    - Any new messages after compaction
  4. Three compaction types exist:
    - Full compaction — API summarizes all old messages
    - Session memory compaction — uses extracted session memory as summary (cheaper)
    - Microcompaction — clears old tool result content when cache is cold (>1h idle)

There are now several comments that (incorrectly?) interpret the undercover mode as only hiding internal information. Excerpts from the actual prompt[0]:

  NEVER include in commit messages or PR descriptions:
  - The phrase "Claude Code" or any mention that you are an AI
  - Co-Authored-By lines or any other attribution

  BAD (never write these):
  - 1-shotted by claude-opus-4-6
  - Generated with Claude Code
  - Co-Authored-By: Claude Opus 4.6 <…>

This very much sounds like it does what it says on the tin, i.e. stays undercover and pretends to be a human. It's especially worrying that the prompt is explicitly written for contributions to public repositories.

[0]: https://github.com/chatgptprojects/claude-code/blob/642c7f94...

The buddy feature the article mentions is planned for release tomorrow, as a sort of April Fools easter egg. It'll roll out gradually over the day for "sustained Twitter buzz" according to the source.

The pet you get is generated based off your account UUID, but the algorithm is right there in the source, and it's deterministic, so you can check ahead of time. Threw together a little app to help, not to brag but I got a legendary ghost https://claudebuddychecker.netlify.app/

> "Anti-distillation: injecting fake tools to poison copycats"

Plot twist: Chinese competitors end up developing real, useful versions of Claude's fake tools.

“Some bullet points are gated on process.env.USER_TYPE === 'ant' — Anthropic employees get stricter/more honest instructions than external use”

Interesting!

I made a visual guide for this https://ccunpacked.dev

The name "Undercover mode" and the line `The phrase "Claude Code" or any mention that you are an AI` sound spooky, but after reading the source my first knee-jerk reaction wouldn't be "this is for pretending to be human" given that the file is largely about hiding Anthropic internal information such as code names. I encourage looking at the source itself in order to draw your conclusions, it's very short: https://github.com/alex000kim/claude-code/blob/main/src/util...

My GitHub fork of anthropics/claude-code just got taken down with a DMCA notice lol

It did not have a copy of the leaked code...

Anthropic thinking 1) they can unring this bell, and 2) removing forks from people who have contributed (well, what little you can contribute to their repo), is ridiculous.

---

DMCA: https://github.com/github/dmca/blob/master/2026/03/2026-03-3...

GitHub's note at the top says: "Note: Because the reported network that contained the allegedly infringing content was larger than one hundred (100) repositories, and the submitter alleged that all or most of the forks were infringing to the same extent as the parent repository, GitHub processed the takedown notice against the entire network of 8.1K repositories, inclusive of the parent repository."

I don't understand the part about undercover mode. How is this different from disabling claude attribution in commits (and optionally telling claude to act human?)

On that note, this article is also pretty obviously AI-generated and it's unfortunate the author didn't clean it up.

I'm amazed at how much of what my past employers would call trade secrets are just being shipped in the source. Including comments that just plainly state the whole business backstory of certain decisions. It's like they discarded all release harnesses and project tracking and just YOLO'd everything into the codebase itself.

Edit: Everyone is responding "comments are good" and I can't tell if any of you actually read TFA or not

> “BQ 2026-03-10: 1,279 sessions had 50+ consecutive failures (up to 3,272) in a single session, wasting ~250K API calls/day globally.”

This is just revealing operational details the agent doesn't need to know to set `MAX_CONSECUTIVE_AUTOCOMPACT_FAILURES = 3`

> Sometimes a regex is the right tool.

I’d argue that in this case, it isn’t. Exhibit 1 (from the earlier thread): https://github.com/anthropics/claude-code/issues/22284. The user reports that this caused their account to be banned: https://news.ycombinator.com/item?id=47588970

Maybe it would be okay as a first filtering step, before doing actual sentiment analysis on the matches. That would at least eliminate obvious false positives (but of course still do nothing about false negatives).

I'd really recommend putting a modicum of work into cleaning up obvious AI generated output. It's rude, otherwise, to the humans you're expecting to read this.

> The multi-agent coordinator mode in coordinatorMode.ts is also worth a look. The whole orchestration algorithm is a prompt, not code.

So much for langchain and langraph!! I mean if Anthropic themselves arent using it and using a prompt then what’s the big deal about langchain

>This was the most-discussed finding in the HN thread. The general reaction: an LLM company using regexes for sentiment analysis is peak irony.

>Is it ironic? Sure. Is it also probably faster and cheaper than running an LLM inference just to figure out if a user is swearing at the tool? Also yes. Sometimes a regex is the right tool.

I'm reading an LLM written write up on an LLM tool that just summarizes HN comments.

I'm so tired man, what the hell are we doing here.

The hooks system is the most underappreciated thing in what leaked. PreToolUse, PostToolUse, session lifecycle, all firing via curl to a local server. Clean enough to build real tooling on top of without fighting it.

The frustration regex is funny but honestly the right call. Running an LLM call just to detect "wtf" would be ridiculous.

KAIROS is what actually caught my attention. An always-on background agent that acts without prompting is a completely different thing from what Claude Code is today. The 15 second blocking budget tells me they actually thought through what it feels like to have something running in the background while you work, which is usually the part nobody gets right.

I am still just shocked that Claude Code was written in Typescript, not C++, Rust or Python.

It also somehow messed up my alacritty config when I first used it. Who knows what other ~/.config files it modifies without warning.

I'm still amazed that something as ubiquitous as "daemon mode" is still unreleased.

- Claude Chat: built like it's 1995, put business logic in the button click() handler. Switch to something else in in the UI and a long running process hard stops. Very Visual Basic shovelware.

- Claude Cowork: same but now we're smarter, if you change the current convo we don't stop the underlying long-running process. 21st century FTW!

- Claude Code: like chat, but in the CLI

- Claude Dispatch: an actual mobile client app, not the whole thing bundled together.

- Daemon mode: proper long-running background process, still unreleased.

>Claude Code also uses Axios for HTTP.

Interesting based on the other news that is out.

I’m more curious how this impacts trust than anything else.

In the span of basically a week, they accidentally leaked Mythos, and then now the entire codebase of CC. All while many people are complaining about their usage limits being consumed quickly.

Individually, each issue is manageable (Because its exciting looking through leaked code). But together, it starts to feel like a pattern.

At some point, I think the question becomes whether people are still comfortable trusting tools like this with their codebases, not just whether any single incident was a mistake.

It is super weird that developers have to run a binary blob on their machines. It's 2026, all the major developer CLI tools are open-source anyway. What's the point for Anthropic to even make it secret?

The irony of ironies is in the last paragraph:

" ...accidentally shipping your source map to npm is the kind of mistake that sounds impossible until you remember that a significant portion of the codebase was probably written by the AI you are shipping.”

Regex for swearing detected, user needs to get more API tokens, he is very very pissed.

Can someone clarify how the signing can't be spoofed (or can it)? If we have the source, can't we just use the key to now sign requests from other clients and pretend they're coming from CC itself?

Can fully AI‑generated code be copyrightable? Is there evidence that the leaked code was AI-generated?

I have yet to see such a company that's so insecure that they would keep their CLI closed source even when the secret sauce is in the model that they control already and is closed source.

Not only that, wouldn't allow other CLIs to be used either.

The "undercover mode" discussion here is exactly the kind of thing non-technical CEOs need to understand — not the implementation, but the governance implication. If your developers are using a tool that actively avoids disclosing its involvement in commits and PRs, your audit trail is broken.

I wrote a short piece explaining the 3 policy implications for teams using Claude Code (or any AI coding tool) — without the technical jargon: https://www.aipolicydesk.com/blog/claude-code-leak-what-ceo-...

The short version: rotate API keys as a precaution, check what audit logs you actually have, and add a clause to your AI policy requiring vendor disclosure of new autonomous capabilities before they get enabled.

It's completely baffling to me why a client that must run on third party environment is behind closed source.

Anyone else have CI checks that source map files are missing from the build folder? Another trick is to grep the build folder for several function/variable names that you expect to be minified away.

Something I’ve been thinking about, somewhat related but also tangential to this topic:

The more code gets generated by AI, won’t that mean taking source code from a company becomes legal? Isn’t it true that works created with generative AI can’t be copyrighted?

I wonder if large companies have throught of this risk. Once a company’s product source code reaches a certain percentage of AI generation it no longer has copyright. Any employee with access can just take it and sell it to someone else, legally, right?

> The obvious concern, raised repeatedly in the HN thread: this means AI-authored commits and PRs from Anthropic employees in open source projects will have no indication that an AI wrote them. It’s one thing to hide internal codenames. It’s another to have the AI actively pretend to be human.

I don’t get it. What does this mean? I can use Claude code now without anyone knowing it is Claude code.

> So I spent my morning reading through the HN comments and leaked source.

> This was one of the first things people noticed in the HN thread.

> The obvious concern, raised repeatedly in the HN thread

> This was the most-discussed finding in the HN thread.

> Several people in the HN thread flagged this

> Some in the HN thread downplayed the leak

when the original HN post is already at the top of the front page...why do we need a separate blogpost that just summarizes the comments?

This is very much AI written, right? The voice sounds like Claude.

As per https://drewdevault.com/2025/04/20/2025-04-20-Tech-sector-re... I congratulate the employee responsible.

I like that if they decide that your usage looks like distillation it just becomes useless, because there’s no way for the end user to distinguish between it just being sort of crappy or sabotaged intentionally. That’s a cool thing to pay for

Sounds like there's still a lot of value in Typescript (otherwise they could have open sourced.)

Plus there's demand for skilled TS software devs that don't ship your company's roadmap using a js.map

20,000 agents and none of them caught it...

> 250,000 wasted API calls per day

How much approximate savings would this actually be?

> Frustration detection via regex (yes, regex)

/\b(wtf|wth|ffs|omfg|shit(ty|tiest)?|dumbass|horrible|awful| piss(ed|ing)? off|piece of (shit|crap|junk)|what the (fuck|hell)| fucking? (broken|useless|terrible|awful|horrible)|fuck you| screw (this|you)|so frustrating|this sucks|damn it)\b/

Personally, I'm generally polite even towards AI and even when frustrated. I simply point out the its mistakes instead of using emotional words.

Absolutely hilarious that it's watching for frustration.

I'd discovered, perhaps mid-2025, that Cursor was noticeably better at fixing bugs if I started cursing at it. Better yet, after a while it would seem to break and start cursing itself ("Oh yes, I see the f*** problem now" and so on). Hilarity ensued.

What a world, where cursing at your machines can make them get their act together.

We're about to reach AGI. One regex at a time...

I'm curious about what people are not looking for about Claude code. What's missing and nobody is talking about? Any clue?

I am curious about these fake tools.

They would either need to lie about consuming the tokens at one point to use in another so the token counting was precise.

But that does not make sense because if someone counted the tokens by capturing the session it would certainly not match what was charged.

Unless they would charge for the fake tools anyway so you never know they were there

They want "Made with Claude Code" on your PRs as a growth marketing strategy. They don't want it on their PRs, so it looks like they're doing something you're not capable of. Well, you are and they have no secret sauce.

The Claude Code leak suggests multi-agent orchestration is largely driven by prompts (e.g., “do not rubber-stamp weak work”), with code handling execution rather than enforcing decisions.

Prompts are not hard constraints—they can be interpreted, deprioritized, or reasoned around, especially as models become more capable.

From what’s visible, there’s no clear evidence of structural governance like voting systems, hard thresholds, or mandatory human escalation. That means control appears to be policy (prompts), not enforcement (code).

This raises the core issue: If governance is “prompts all the way down,” it’s not true governance—it’s guidance.

And as model capability increases, that kind of governance doesn’t get stronger—it becomes easier to bypass without structural constraints.

Has anyone actually implemented structural governance for agent swarms — voting logic, hard thresholds, REQUIRES_HUMAN as architecture not instruction?

I'm surprised that they don't just keep the various prompts, which are arguably their "secret sauce", hidden server side. Almost like their backend and frontend engineers don't talk to each other.

> Anti-distillation: injecting fake tools to poison copycats

Does this mean `huggingface.co/Jackrong/Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled` is unusable? Had anyone seen fake tool calls working with this model?

Come on guys. Yet another article distilling the HN discussion in the original post, in the same order the comments appear in that discussion? Here's another since y'all love this stuff: https://venturebeat.com/technology/claude-codes-source-code-...

The irony of an IP scraper on an absolutely breathtaking, epic scale getting its secret sauce "scraped" - because the whole app is vibe coded (and the vibe coders appear to be oblivious to things like code obfuscation cuz move fast!)...

And so now the copy cats can ofc claim this is totally not a copy at all, it's actually Opus. No license violation, no siree!

It's fucking hilarious is what it is, it's just too much.

I used to swear at Claude. To be honest, I thought it helped get results (maybe this is "oldschool" LLM thinking), but I realized it was just making me annoyed.

> “Do not rubber-stamp weak work” and “You must understand findings before directing follow-up work. Never hand off understanding to another worker.”

:-D

Guys I’m somewhat suspicious of all the leaks from Anthropic and think it may be intentional. Remember the leaked blog about Mythos?

What a cesspool. So this is the power of being 80x more productive, having infinite llm usage quota? No wonder they had to let Satan take the wheel and went 100% vibe code. Thanks for making a point, llms are a disgrace

Why didn't they open the source themselves? What's the point of all this secrecy anyway?

Claude Code’s Source Code Leaked Through npm. Here’s What Actually Happened. https://www.everydev.ai/p/tool-claude-codes-source-code-leak...

What effect will this have on their IPO? Can someone take the code and make a clone?

A few weeks ago I was using Opus and Sonnet in OpenCode. Is this not possible anymore?

> As one Twitter reply put it: “accidentally shipping your source map to npm is the kind of mistake that sounds impossible until you remember that a significant portion of the codebase was probably written by the AI you are shipping.”

To err is human. AI is trained on human content. Hence, to err is AI. The day it stops making mistakes will be the beginning of the end. That would mean the existence of a consciousness that has no weakness. Great if it’s on your side. Terrible otherwise.

how sure are we this entire “accident” is not an aprils fools joke??

Genius level AI marketing

I wrote this an hour ago and it seems that Claude might not understand it as frustration:

> change the code!!!! The previous comment was NOT ABOUT THE DESCRIPTION!!!!!!! Add to the {implementation}!!!!! This IS controlled BY CODE. *YOU* _MUST_ CHANGE THE CODE!!!!!!!!!!!

Built a tool to ask questions on the Claude Code source code: https://askgithub.com/alex000kim/claude-code

Meanwhile Claude Code is still awesome. I don’t see my self switching to OpenAI (seriously bad mgmt and possibly the first domino to fall if there is a correction) or Gemini (Google ethics cough cough).

I hope cheap Chinese models will overtake Anthropic.

Can someone ask claude to write a deep dive on how compaction works and why it’s so slow? (I still can’t fathom why they wouldn’t just add a user message “compact the conversation we’ve just had”

that frustration regex is missing "idiot", which is the most common frustration word I use with code-agents

Does this mean I can now self host Claude?

that’s fun am having exploring this codebase with claude code, inception at its best

But what does Claude do when it detects user fruatration?! Don't leave us hanging here!

Edit: it gets sent to Anthropic via telemetry and it ends up on the fuck chart!

https://old.reddit.com/r/ClaudeCode/comments/1s99wz4/boris_t...

Undercover mode is the most concerning part here tbh.

re: binary attestation: "Whether the server rejects that outright or just logs it is an open question"

...what we did at Snap was just wait for 8-24 hours before acting on a signal, so as not to provide an oracle to attackers. Much harder to figure out what you did that caused the system to eventually block your account if it doesn't happen in real-time.

(Snap's binary attestation is at least a decade ahead of this, fwiw)

"and i also wrote this using claude" -- can we just include that at this point?

Probably an unpopular opinion but Anthropic are too popular for their own good.

1. They are loved, and for good reasons, Sonnet 4 was groundbreaking but Opus 4.6 was for many a turning point in realizing Agentic SDLC real potential. People moved from Cursor to Claude Code in droves, they loved the CLI approach (me too), and the LOVED the subsidized $200 max pro plan (what's not to love, pay $200 instead of $5000 to Cursor...) They are the underdog, the true alternative to "evil" OpenAI or "don't be evil" Google, really standing up against mass surveillance or use of AI for autonomous killing machines. They are standing for the little guy, they are the "what OpenAI should have been" (plus they have better models...) They are the Apple of the AI era.

2. They are too loved, so loved that it protects them from legitimate criticism. They make GitHub's status page look good, and they make comcast customer service look like Amazon's. (At least Comcast has customer service), They are "If Dario shoots a customer in the middle of 5th avenue it won't hurt their sales one bit" level of liked. The fact they have the best models (for now) might be their achilles heel, because it hides other issues that might be in the blindspot. And as soon as a better model comes out from a competitor (and it could happen... if you recall OpenAI were the undisputed kinds with GPT 4o for a bit) these will become much more obvious.

3. This can hurt them in the long run. Eventually you can't sustain a business where you have not even 2 9s of SLA, can't handle customer support or sales (either with humans or worse for them - if they can't handle this with AI how do they expect to sell their own dream where AI does everything?). I'm sure they'll figure it out, they have huge growth and these are growth pains, but at some point, if they don't catch up with demand, the demand won't stay there forever the moment OpenAI/Google/someone else release a better model.

4. They inadvertently made all of the cybersecurity sector a potential enemy. Yes, all of them use Anthropic models, and probably many of them use Claude Code, but they know they might be paying the bills of their biggest competitor. Their shares drop whenever Anthropic even hints of a new model. Investors cut their valuations because they worry Anthropic will eat them for breakfast. I don't know about you, but if you ask me, having the people who live and breath security indirectly threatened by you, is not the best thing in the world, especially when your source code is out in the open for them to poke holes in...

5. the SaaS pocalypse - many of Claude Code's customers are... SaaS companies, that the same AI is "going to kill", again, if there was another provider that showed a bit more care about the entire businesses it's going to devour, if they also had even marginally better models... would the brand loyalty stay?

Side note: I'm an Claude Enterprise customer, I can't get a human to respond to anything, even using the special "enterprise support" methods, and I'm not the only one, I know people who can't get a sales person, not to mention support, to buy 150 + seats (Anthropic's answer was - release self serve enterprise onboarding, which by the way is "pay us $20 which does not include usage, usage is at market prices, same as getting an API key", you pay for convenience and governance, p.s. you can't cancel enterprise, it's 20 seats min, for 1 year, in advance, so make sure you really need it, the team plan is great for most cases but it lacks the $200 plan, only the $100 5x plan).

[dead]

[dead]

[dead]

[dead]

[flagged]

[dead]

[dead]

[flagged]

[dead]

[dead]

[flagged]

[dead]

[dead]

[flagged]

[flagged]

[dead]

[dead]

[dead]

[dead]

[dead]

[flagged]

[dead]

[dead]

[dead]

[dead]

[dead]

[dead]

[dead]

[flagged]

why claude bring back 4o u dumb fks

Can somebody tell me what this means for the company?

The feature flag names alone are more revealing than the code. KAIROS, the anti-distillation flags, model codenames those are product strategy decisions that competitors can now plan around. You can refactor code in a week. You can't un-leak a roadmap.