He Built a Million-User Anonymous Social Viewer with Claude Code — Then Meta Deleted His Whole Family's Accounts

A developer with over 7 years of experience used Claude Code to help build an anonymous social media viewer called Spybroski, which grew to roughly 1 million active users within a year. Then Meta not only banned his long-dormant Facebook account, but also deleted his father's, mother's, and sister's accounts, and even shut down his family business's page and ad account. The incident sparked heated discussion on Reddit and raised a sharp question: as AI dramatically lowers the barrier to reverse engineering and tool development, will big platforms respond by respecting privacy more — or by punishing harder?

What Is Spybroski: Viewing Public Content Without Logging In or Leaving a Trace

Spybroski's core function is straightforward: it lets users view public content on Instagram, TikTok, Snapchat, and other platforms without registering or logging into any social account — and without appearing in the other person's story view history. The site offers a free trial with no registration required; users just open it, pick a platform, and enter a username.

The developer says his target users are people who don't have social media accounts or don't want to be tracked. That need sounds niche, but the growth numbers suggest otherwise — after posting in a few subreddits, the product reached roughly 1 million active users within a year. According to the original Reddit post, this growth came almost entirely from community word-of-mouth rather than paid advertising.

From a product standpoint, Spybroski hit a long-standing gray area: public content is visible in itself, but platforms turn "viewing" into an act with a privacy cost through login walls, view histories, and tracking mechanisms. Spybroski tries to separate the two.

Login walls and view tracking are key to understanding this type of tool. A login wall refers to platforms requiring users to register and log in before viewing content — Instagram, for example, shows only a tiny number of posts to logged-out visitors before popping up a login prompt. The goal is to convert anonymous browsing into identifiable, trackable account behavior. As for story view history, both Instagram and Snapchat show the poster who viewed their ephemeral content, ordered chronologically. This means that simply "viewing" leaves a social trace that can expose a user's interest in someone. Tools like Spybroski work by proxying requests server-side to scrape public data, then presenting it to the user through a third-party server, so the platform's tracking mechanisms can't link the browsing behavior to a specific individual. This is precisely the root of its conflict with platform terms of service.

What Claude Code Helped With: Parsers Went from Days to Hours

The developer has over 7 years of web scraping experience, but this time he deeply integrated Claude Code into his development workflow. The core use case was having Claude work directly in his codebase, writing the backend logic to scrape and parse public profiles, posts, and stories for each platform.

Each platform's data structure is different, which is the most time-consuming part of scraping work. He noted that parsing work that used to take days was shortened to hours with Claude Code. Claude excels at "getting code to run," especially when handling highly repetitive parsing tasks with varying patterns.

But the other side of the story matters just as much: when traffic surged, the code couldn't hold up. The developer had to step in personally and redo the performance-side implementation. His conclusion was blunt — Claude is great at "making it work," but "making it handle a million users" still requires a human. This is a sober footnote on the current boundaries of AI coding tools: AI can dramatically accelerate prototyping and feature implementation, but scaling architecture, performance tuning, and stability still depend on engineering judgment.

Web scraping and parsers refer to an automated pipeline: a program simulates browser requests to a target page, retrieves the returned HTML or JSON data, and extracts structured fields like usernames, post content, and timestamps. A "parser" is the module specifically responsible for mapping messy raw data into tidy fields. Because each platform's page structure, anti-scraping strategies, and data interface formats vary widely and change frequently, parsers often need to be rewritten repeatedly — making them the most time-consuming and failure-prone part of scraping work. Claude Code can significantly speed things up because it can read existing implementations directly in the codebase, infer target fields, and generate corresponding parsing logic, compressing repetitive manual field-by-field debugging into conversational iteration. But a parser that "works" isn't the same as one that "holds up" — under high concurrency, engineering measures like caching, queues, rate limiting, and proxy pools are needed to ensure stability.

Meta's Counterattack: From Personal Accounts to Family-Wide Punishment

The real turning point came after Meta noticed Spybroski. The developer claims Meta somehow dug up his years-unused old Facebook account and then launched a series of bans:

• His father's Facebook account was deleted
• His mother's Facebook account was deleted
• His sister's Facebook account was deleted
• The family business page and ad management account were deleted
• The person who designed the family's website received a notice

He also received notices via email and physical mail, with letters sent directly to his residence. The developer emphasized that his family has nothing to do with Spybroski — their only "crime" is sharing his last name.

If this account is accurate, it means the platform's enforcement action has gone beyond the product itself to affect relatives and business entities who had no involvement in development. For a company whose business model is built on tracking users, this kind of "guilt by association" approach is itself a strong signal.

The linked-account mechanism in platform bans deserves separate explanation. Terms of service at Meta and similar platforms typically allow them to take action based on "associated accounts," judged by factors like the same device fingerprint, identical IP ranges, shared payment methods, similar names, or historical interaction relationships. This mechanism was originally designed to combat fake accounts, fraud networks, and organized violations, but in practice it can sweep in relatives, colleagues, or business entities unrelated to the violation. The deletion of the family business page and ad account is especially severe, since ad accounts are often tied to real payment information, and bans directly impact business revenue. The developer receiving physical mail indicates the platform invoked legal notice or formal warning channels, not just in-app prompts. This "guilt by association" enforcement sits in a controversial compliance zone: it expands the scope of accountability from specific actions to social relationships, with deterrence effects and collateral damage risks equally significant.

An Irony: A Company Built on Tracking Is Furious About an Anti-Tracking Tool

The developer himself pointed out the irony in his post: a company whose entire business model is built on tracking users is furious about a tool that helps people avoid being tracked.

This contradiction isn't new, but AI is changing its scale. In the past, reverse engineering and scraping were the domain of a few experts; now, a developer with a laptop and tools like Claude can build in weeks what used to require an entire team. The developer believes big platforms understand this, and their response isn't stronger privacy protections but punishment — deleting accounts, deleting family members, deleting businesses — making an example to force the next person who wants to build something similar to think twice.

He also judges that this strategy won't work. More similar tools are appearing, and the box can't be closed. What really worries him is how far platforms are willing to go, and who else will get caught up in it.

Implications for Developers and Platform Governance

The incident sparked discussion in the tech community, with several aspects worth noting.

First, AI coding tools are democratizing reverse engineering and scraping capabilities. Claude Code's efficiency gains in parser development are direct evidence of this trend. One person with a laptop can indeed build products that used to require a team.

Second, scaling remains the moat for human engineers. The developer explicitly stated that Claude can write features, but the performance overhaul to handle a million users has to be done by himself. This provides a concrete counterexample to the narrative that "AI will replace engineers."

Third, the boundary between platform enforcement and family-wide punishment deserves scrutiny. If the product itself sits in a gray area of law and platform terms, banning the developer's account is understandable; but deleting accounts of family members who weren't involved in development and family business assets expands accountability to associations rather than actions.

Fourth, the demand for privacy tools is real and substantial. One million active users is a number that shouldn't be ignored. When platforms use login walls and tracking mechanisms to increase the cost of viewing, tools that bypass these mechanisms will keep appearing.

The developer said in closing that he's not sure how far platforms will go, or who will be caught up next. This question currently has no answer, but Spybroski's experience at least shows that after AI lowered development barriers, the conflict between platforms and developers is entering a new phase.

The democratization of reverse engineering is the broader context of this incident. Traditionally, reverse engineering required understanding network protocols, cryptographic signatures, obfuscated code, and anti-scraping countermeasures — the expertise of a few security researchers and senior scraping engineers. As large models like Claude can directly read codebases, explain interface logic, and generate working implementations, this barrier has been dramatically lowered. The result is that individual developers can replicate in weeks what used to be team-level workloads, and tools like Spybroski will become more common. Platforms therefore face not isolated violators but a continuously emerging long-tail supply. Simply banning a single developer or associated accounts is unlikely to change the economics of the supply side; what would truly reduce demand for such tools is lowering the privacy cost of legitimate viewing — for example, offering more transparent public browsing options or more restrained tracking mechanisms — otherwise governance will remain stuck in a game of whack-a-mole.

One-line Summary

Developer builds million-user anonymous social viewer with Claude Code, gets whole family banned by Meta in sweeping crackdown.

Paragraph Summary

A developer with 7+ years of experience used Claude Code to build Spybroski, an anonymous social viewer letting users browse Instagram, TikTok, and Snapchat public content without logging in or leaving view traces, reaching ~1M active users in a year. Claude Code mainly accelerated multi-platform parser development, cutting days of work to hours, though performance overhaul for a million users still required human effort. Meta then banned the developer's old Facebook account and deleted his parents' and sister's accounts plus family business page and ad account, highlighting how AI is lowering reverse-engineering barriers while raising questions about platform enforcement boundaries.