evalstate

Extract and analyze writing improvements from GitHub PR review comments. Use when asked to show review feedback, style changes, or editorial improvements from a GitHub pull request URL. Handles both explicit suggestions and plain text feedback. Produces structured output comparing original phrasing with reviewer suggestions to help refine future writing.

Investigate fast-agent session and history files to diagnose issues. Use when a session ended unexpectedly, when debugging tool loops, when correlating sub-agent traces with main sessions, or when analyzing conversation flow and timing. Covers session.json metadata, history JSON format, message structure, tool call/result correlation, and common failure patterns.

Use this skill when the user wants a polished, shareable, source-grounded, self-contained HTML artifact in the Birch visual system for engineering, product, planning, review, status, incident, process, codebase, data, or benchmark communication. Trigger for one-pagers, visual summaries, reports, dashboards, module/process explainers, implementation plans, design-direction comparisons, PR/change writeups, findings-first code reviews, incident/status reports, briefing decks, flow diagrams, numeric/data briefs, model/run/token/ time/cost reports, and design-system references. Do not use it for editing production web apps, React/Vue components, or ordinary chat answers.

This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.