effort-estimation
Calibrate engineering effort estimates for git commits using a 5-tier rubric. Use whenever you need to translate a diff into an hours estimate for a senior engineer working with or without an AI coding assistant.
A meta-repository for Claude Code users that includes workspace setup (skills, commands, hooks) and an evaluator for your Claude Code configuration
Deep-evaluate a single skill â static analysis, contextual rubric scoring, and A/B redundancy testing. Runs all 3 layers on one skill to determine if it earns its place.
capture-feedback
Capture structured thumbs up/down feedback with context, tags, and optional rubric scores after completing a task.
agent-project-development
This skill should be used when the user asks to "start an LLM project", "design batch pipeline", "evaluate task-model fit", "structure agent project", or mentions pipeline architecture, agent-assisted development, cost estimation, or choosing between LLM and traditional approaches. NOT for evaluating agent quality or building evaluation rubrics (use agent-evaluation), NOT for multi-agent coordination or agent handoffs (use multi-agent-patterns).
autocontext
Iterative strategy generation and evaluation system. Use when the user wants to evaluate agent output quality, run improvement loops, queue tasks for background evaluation, check run status, or discover available scenarios. Provides LLM-based judging with rubric-driven scoring.
email-design-review
Blind, adversarial review of a finished email design in Figma, scored against the brand's EMAIL.md and a weighted rubric before anyone builds it. The critic role: the deliverable is a scorecard and a short list of fixes, never edits to the frame. Use after an email is designed and before it is built in the email platform, or whenever someone asks to review, QA, grade, check or score an email design. Run it as a separate agent that never saw the build.
article-optimizer
Use when the user wants to improve a specific article under a fixed viral-scoring rubric, iterate on one article to raise its score, or run single-article scoring plus rewrite loops. Best for content optimization experiments where the scoring prompt stays fixed and the article text is the thing being changed.
assessment-design
Evidence-based assessment design with rubrics, feedback strategies, and formative checkpoints. Aligns each assessment to learning objectives using Bloom's taxonomy. Applies Nicol's 7 principles of good feedback practice. Reads from /learning-objectives manifest and extends it with assessment specs. (idstack)
ai-evals
Help users create and run AI evaluations. Use when someone is building evals for LLM products, measuring model quality, creating test cases, designing rubrics, or trying to systematically measure AI output quality.
cvfit
Use when someone points at a job, hackathon, fellowship or program application and wants a resume for it. Triggers: a posting URL, a screenshot of a posting, pasted posting text, a company plus role name, or phrasings like 'adapt my CV to this', 'tailor my resume for this role', 'Harvard format resume', 'make a CV for this application', 'one-page resume'. The skill reads the posting, assembles the person's material from whatever exists (the conversation, an Obsidian vault, markdown notes, an old resume, public repos), decides the right shape for that person and that posting, renders a one-page Harvard-format PDF, and evaluates it against a rubric until it passes or the gap is explained. Don't use for cover letters, LinkedIn profiles or portfolio sites.
multi-model-analysis
Runs 3 AI models in parallel to independently analyze a problem and propose approaches. Compares all proposals against a rubric and selects the best. Can be invoked standalone or called from any agent (issue-resolver, pr-review, etc.).
agents-md-improver
Audit and improve project-rules files (AGENTS.md, CLAUDE.md, .agents/instructions, local overrides) so the agent keeps accurate project context. Use when the user asks to check, audit, review, update, improve, or fix their AGENTS.md or CLAUDE.md, mentions "project rules maintenance" or "agent context optimization", or when the codebase has changed enough that the rules file may be stale. Scans the repository for every rules file, grades each against a quality rubric, outputs a quality report, and applies targeted edits only after user approval.