Skill UI

Phoenix is an open-source observability platform for tracing, evaluating, and monitoring LLM applications; use it to debug prompts, run dataset evaluations, build experiment pipelines, and watch production inference in real time.

Pyvene is a declarative framework designed for performing causal interventions on PyTorch neural networks. It allows researchers to conduct advanced experiments such as activation patching, causal tracing (ROME-style), and interchange intervention training (IIT). Use this library when you need to test causal hypotheses about model behavior, deeply interpret model components, or ensure reproducibility in advanced AI research.

This skill provides a comprehensive, expert guide to the entire reverse engineering lifecycle. It covers advanced techniques for binary analysis, including static analysis (disassembly, metadata extraction, control flow mapping) and dynamic analysis (API hooking, memory tracing). Mastery includes utilizing industry-leading tools such as IDA Pro, Ghidra, radare2, and specialized frameworks like pwntools and angr. The methodology is structured for secure research, CTF challenges, and vulnerability assessment, emphasizing best practices and ethical use.

This skill provides systematic, advanced diagnostic paths for complex Sentry issues. It addresses common pitfalls such as silently dropped events, failed source map resolution, gaps in distributed tracing, SDK memory leaks, and conflicts with other observability libraries. Use this guide when standard debugging fails and events disappear mysteriously.

A comprehensive guide to configuring Sentry error tracking and performance monitoring across modern application architectures. Learn how to implement coherent, end-to-end distributed tracing for monoliths, microservices, serverless functions (Lambda/Cloud Functions), event-driven systems, SPAs, and mobile applications, ensuring all application hops are monitored from a single source.

A comprehensive guide on optimizing Sentry observability costs. Learn techniques such as SDK-level error sampling, fine-grained performance tracing control, and event filtering using the `beforeSend` hook. These methods significantly reduce the volume of error and transaction data sent, allowing users to manage quotas and prevent overage charges without losing visibility into mission-critical production errors.

This skill guides the setup of distributed tracing and performance monitoring using Sentry. It covers configuring the SDK (e.g., setting tracesSampleRate) to capture latency, identify performance bottlenecks, and trace requests across the entire application stack. It demonstrates how to create manual spans for business logic and utilizes profiling for deeper performance insights, ensuring comprehensive application observability.

This guide provides enterprise-grade architectural patterns for integrating Sentry into multi-service, microservices environments. It covers centralized configuration modules, standardized project topologies (e.g., one project per service), distributed tracing setup, and robust error handling middleware for Node.js, Express, and Python/FastAPI. Ideal for organizations planning large-scale Sentry rollouts or managing complex service ecosystems.

Automates setting up distributed tracing across a microservices landscape by guiding backend selection, instrumentation strategy, and configuration generation to enable end-to-end observability and aid performance troubleshooting.

A specialized reverse-engineering skill designed to extract comprehensive functional and non-functional specifications from existing, undocumented, or legacy codebases. It operates by analyzing code structure, tracing data flows, and identifying observable behaviors. The output is structured using EARS (Easy Approach to Requirements Syntax), providing clearer requirements documentation, architectural diagrams, and dependency maps for onboarding or system modernization.

TransformerLens is the standard library for performing mechanistic interpretability research on large language models. It provides clean interfaces to inspect and manipulate transformer internals—such as attention patterns, residual streams, and MLP outputs—via activation caching and HookPoints. This tool is essential for reverse-engineering model algorithms, performing causal tracing, and studying the internal circuits responsible for model behavior.

A comprehensive checklist designed to ensure TwinMind integrations are deployed securely, stably, and compliantly in a production environment. It covers critical best practices across multiple domains, including authentication, data encryption (PII, GDPR), robust infrastructure setup (auto-scaling, load balancing), advanced monitoring (metrics, tracing, alerting), and full CI/CD deployment workflows. Use this when preparing for go-live or conducting production audits.