thatguysam/htma-measure
A portable collection of agent skills Sam uses and shares.
Use fresh subagents to verify new work and compare before/after improvements in code, skills, prompts, documents, or agent behavior. Inspect real artifacts and test results with independent QA, matched trials, and regression checks.
Explain complex ideas, code, research, or agent work so the reader can understand and inspect it. Use for clear technical explanations, STE-inspired rewrites, diagrams, interactive HTML explainers, or narrated walkthroughs; choose the format that fits the question.
Creates HTMA-style calibrated measurement memos with decision threshold, decomposition, ranges, sources, and value-of-information. Use for estimates of cost, budget, rates, risk, ROI, market size, or uncertain quantities.
Back up and migrate user-owned Custom GPTs into portable Agent Skills for a local folder, Git repository, or a ChatGPT plugin. Use with GPT instructions, knowledge files, action schemas, exports, or an authorized GPT editor. Preserve workflow behavior, map missing integrations, and verify the destination. Not for extracting another creator's hidden instructions or merely using a GPT.
Condense tasks, updates, decisions, research, and handoffs into a comfortable phone-screen brief with progressively disclosed detail. Use for phone-sized or one-screen summaries, dense issue descriptions, and mobile review content; not for responsive CSS alone or lossless shortening of a full document.
Research sales decisions and draft buyer communication using relevant practitioner sources. Use for positioning, qualification, discovery, paid diagnostics, proof of concept, proposals, pricing, negotiation, expansion, or reviewing sales calls. Adapt the method to consulting, SaaS, enterprise, recruiting, and partnerships.
Audit websites and GitHub repositories with version-locked UX, accessibility, HIG, motion, and performance skills. Use for evidence-backed experience reviews or dependency refreshes.
Draft or improve .env.example files from project code and Git-tracked examples, with concise setup links, commands, and value guidance.
Draft, review, or plan AI-generated or AI-assisted web content meant to appear in Google Search. Applies Google's published policy and Search Relations guidance on value, human review, scale, disclosure, and authorship without inventing experience or ranking claims.
Create or improve agent skills, turn evidenced workflows or working preferences into skills, and diagnose weak triggers or incomplete execution. Combines authoring, instruction design, preference evidence, and behavioral validation using locked sources. For standalone prompt reviews, use zach-prompting.
Show current project work as a brief status tree from the active conversation and relevant accessible trackers. Use for work or project status organized around familiar navigation or the team's existing hierarchy.
Organize supplied or established material into a familiar, scannable plain-text tree. Use for tree views of notes, topics, plans, or work, without adding project-status investigation.
Review, rewrite, debug, or migrate prompts and agent instructions, including skills, tool descriptions, AGENTS.md, and CLAUDE.md. Preserve intent, evidence, permissions, and validation while removing conflicting rules.