Skill UI

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.

This skill wraps the `getCompactContext` functionality to efficiently retrieve a compacted ReasoningBank context, specifically optimized for cost-analysis queries. It generates a detailed report summarizing critical metrics such as tokens saved, memories retrieved, and agentic-flow availability. It is essential for monitoring and optimizing resource usage in complex AI workflows, providing deep visibility into the efficiency and cost of the context retrieval mechanism.

This skill provides a granular, session-by-session view of total accumulated costs. It lists metrics such as start time, message count, top model, and total cost for every recorded conversation. This is ideal for identifying which specific sessions consumed the most resources, which is crucial for project retrospectives and evaluating the sustainability of current session patterns.

Performs multi-baseline counterfactual cost analysis by comparing actual session spending against hypothetical costs generated by fixed, non-optimal models (e.g., always-haiku, always-sonnet). This tool quantifies the financial savings realized by the smart routing mechanism, answering the critical question: 'Is the routing earning its keep?' It is ideal for quarterly cost reviews, setting CI performance gates, and validating routing configurations.

This tool detects cost regressions by comparing two cost-summary JSON snapshots (baseline vs. current). It is essential for PR-gate workflows, checking for unexpected spending increases using three orthogonal signals: total percentage change, absolute dollar increase, and disproportionate growth in specific token classes (e.g., cache writes).