Skill UI

Implements a robust caching strategy by using the SHA-256 hash of file content as the primary key, instead of the file path. This pattern is essential for high-cost file processing pipelines (like OCR, PDF parsing) because it ensures cache hits survive file renaming or moving, while still automatically invalidating entries when the content changes. It adheres to the Single Responsibility Principle (SRP) by isolating caching logic into a service layer wrapper.

This tool audits the token consumption across all components (agents, skills, rules, MCP servers) within a Claude Code session. It analyzes token overhead, identifies bloat (e.g., overly long descriptions, redundant components), and generates a detailed context budget report. Use it to maintain optimal context window usage, improve performance, and prevent context saturation.

This pattern defines a robust, continuous loop structure for autonomous agents. It integrates crucial components like quality gates, evaluation harnesses, and explicit recovery controls. It guides complex workflows by defining decision paths (e.g., PR control, RFC decomposition) and includes strategies to mitigate common failure modes such as infinite loop churn or cost drift, ensuring measurable progress in autonomous systems.

This skill analyzes completed code sessions to automatically extract reusable patterns, transforming raw interaction data into structured, learned skills. It allows the AI to continuously improve its domain knowledge by capturing techniques like error resolution, debugging methods, and workarounds, making it effective for advanced coding assistance and knowledge curation.

An advanced system that analyzes coding sessions via hooks to extract atomic, confidence-scored behaviors called "instincts." It tracks patterns, allowing users to evolve these instincts into comprehensive skills, commands, or agents. V2.1 introduces project-scoped learning, preventing knowledge contamination between different codebases and improving development workflow efficiency.

This tool performs robust outlier detection on individual session spending. Utilizing Median and Median Absolute Deviation (MAD), it calculates a modified z-score, making it highly resistant to extreme values, unlike mean+sigma methods. It specifically identifies which singular sessions are anomalous, making it crucial for cost monitoring, auditing, and integrating into CI/CD pipelines to prevent unexpected resource overruns.

A comprehensive pattern for building robust and cost-efficient LLM application pipelines. This guide details combining model routing based on task complexity, immutable cost tracking, limited retry logic, and prompt caching to manage API expenditures while maintaining high quality. Essential for production systems handling variable-complexity, high-volume tasks.

This skill runs comprehensive benchmarks against structural and adversarial corpora. It measures the performance and cost efficiency of a booster model compared to leading LLM baselines (Gemini, Anthropic). Results are persisted in verifiable JSON reports, crucial for auditing claims and verifying booster win rates before publishing.

A high-performance utility for executing simple, deterministic code transformations (e.g., removing console logs, converting var to const). This engine leverages a WASM core to perform structural refactoring sub-millisecond, without calling any expensive LLM APIs. Ideal for CI/CD pipelines, automated bulk refactoring, and cost-sensitive development workflows where determinism and speed are paramount.

This tool routes tasks through a multi-tier system, specifically partitioning tasks into Agent Booster eligible (Tier 1) and standard LLM calls (Tier 2/3). It calculates and reports the estimated cost, leveraging the structural $0 cost of Booster bypasses for simple transforms, helping users optimize spend before running batches of similar tasks.

This skill monitors operational costs against defined budgets by reading accumulated spending and computing utilization. It emits a detailed 4-tier alert ladder (INFO, WARNING, CRITICAL, HARD_STOP) based on predefined thresholds (50%, 75%, 90%, 100%). It is essential for cost governance, ensuring resources are managed within budget limits, and preventing system overspending.

Monitors the acceleration trend of production spending (burn rate) over defined time windows. It compares the latest spending bucket against the historical average, providing an early warning system. This is crucial for detecting sudden, unexpected cost spikes, such as a 'hot loop,' independent of overall budget limits.