Back to List
Anthropic Claude Code Leak Reveals TypeScript Source Map and Experimental Always-On Agent Features
Industry NewsAnthropicCybersecurityArtificial Intelligence

Anthropic Claude Code Leak Reveals TypeScript Source Map and Experimental Always-On Agent Features

A significant data leak has impacted Anthropic following the release of the Claude Code 2.1.88 update. Users discovered that the update inadvertently included a package with a source map file containing the tool's TypeScript codebase. The leak, which was highlighted by a user on X (formerly Twitter), reportedly exposes over 512,000 lines of code. This accidental disclosure provides an unprecedented look into the internal mechanics of Claude Code, including references to a Tamagotchi-style 'pet' feature and an always-on agent. The incident has sparked intense discussion within the developer community as the source code provides a direct window into Anthropic's development process and upcoming experimental features that were not yet intended for public scrutiny.

The Verge
Share

Key Takeaways

  • Anthropic's Claude Code 2.1.88 update accidentally included a source map file revealing its TypeScript codebase.
  • The leak reportedly consists of more than 512,000 lines of code.
  • Discovered features within the code include a Tamagotchi-style 'pet' and an always-on agent functionality.
  • The exposure was first brought to public attention by a user on the social media platform X.

In-Depth Analysis

The Nature of the Claude Code Leak

The leak occurred immediately following the deployment of the 2.1.88 update for Claude Code. Unlike standard software releases where code is obfuscated or compiled, this specific update contained a package with a source map file. In software development, source maps are typically used for debugging, as they map transformed code back to the original source. By including this file, Anthropic inadvertently allowed users to reconstruct and view the original TypeScript codebase, leading to the exposure of over 512,000 lines of internal logic.

Discovery of Experimental Features

As developers and researchers parsed through the leaked data, two specific elements caught significant attention: a Tamagotchi-style 'pet' and an 'always-on' agent. While the original news does not detail the specific functionality of these features, their presence in the codebase suggests that Anthropic has been experimenting with more interactive and persistent AI behaviors. The 'pet' concept implies a gamified or personality-driven interface, while the 'always-on' agent suggests a shift toward autonomous AI that operates continuously rather than just in response to specific prompts.

Industry Impact

This incident highlights the ongoing security and privacy challenges faced by AI companies during rapid deployment cycles. The exposure of a massive TypeScript codebase provides competitors and researchers with a blueprint of Anthropic’s engineering approach. Furthermore, the leak of unreleased features like the 'always-on' agent may force Anthropic to accelerate its public roadmap or address safety concerns regarding persistent AI agents earlier than planned. For the broader industry, it serves as a cautionary tale regarding the inclusion of source maps in production-level updates.

Frequently Asked Questions

Question: How did the Claude Code leak happen?

The leak occurred because the 2.1.88 update for Claude Code included a source map file that contained the tool's TypeScript codebase, which was not intended for public release.

Question: What specific features were found in the leaked code?

The leaked code reportedly contains references to a Tamagotchi-style 'pet' and an always-on agent, indicating new directions for the Claude interface and functionality.

Question: How much code was actually exposed?

According to reports from users on X, the leaked data contains more than 512,000 lines of code.

Read Original Article

Related News

Meituan Showcases AI Innovations at ACL 2026: From Model Evaluation to Reasoning Optimization and Generative Paradigms
Industry News

Meituan Showcases AI Innovations at ACL 2026: From Model Evaluation to Reasoning Optimization and Generative Paradigms

Meituan's technical team has announced the acceptance of six research papers at ACL 2026, a premier international conference in computational linguistics and natural language processing. The papers cover a broad spectrum of cutting-edge AI fields, including large model evaluation, complex process reasoning, and competition-level mathematical thinking optimization. Additionally, the research explores advancements in reinforcement learning and generative recommendation systems. These contributions signify Meituan's strategic focus on building a new paradigm for generative AI, aiming to enhance the logical depth and practical utility of language models. By addressing both theoretical benchmarks and real-world application challenges, Meituan continues to position itself at the forefront of NLP research, contributing to the evolution of how AI systems reason, learn, and interact with users in complex environments.

Meituan LongCat Team Launches General 365: A New Benchmark Revealing Critical Gaps in AI Reasoning Capabilities
Industry News

Meituan LongCat Team Launches General 365: A New Benchmark Revealing Critical Gaps in AI Reasoning Capabilities

The Meituan LongCat team has officially released General 365, a rigorous new benchmark designed to evaluate the reasoning capabilities of modern artificial intelligence. In an initial assessment of 26 mainstream models, the results reveal a significant performance gap across the industry. Even Gemini 3 Pro, currently identified as the most powerful model in the test, achieved an accuracy rate of only 62.8%. Furthermore, the vast majority of the models tested failed to reach the 60% threshold, which is traditionally considered a passing grade. This release by Meituan's technical team establishes a new standard for measuring logical depth in AI and highlights the substantial room for improvement in complex reasoning tasks.

Managing AI Coding with Agent Evaluation: Meituan's Practice in Refactoring 310,000 Lines of Code
Industry News

Managing AI Coding with Agent Evaluation: Meituan's Practice in Refactoring 310,000 Lines of Code

Meituan's technical team has introduced a groundbreaking approach to managing AI-assisted development, focusing on the refactoring of 310,000 lines of code. As AI now generates over 90% of code in certain environments, the primary challenge has shifted from production speed to the management of AI's output quality. The team argues that without unified standards, AI can exponentially increase technical debt and system chaos. To combat this, Meituan implemented an 'Agent evaluation' mindset, utilizing four key pillars: technical debt sorting, rule construction, a standardized Refactoring SOP, and a Pre-PR (Pull Request) mechanism. This strategy successfully transitions code refactoring from a high-cost, specialized project into a sustainable, daily iterative process, ensuring long-term system stability in the era of AI-dominated coding.