An anti-AI-slop writing standard and checker for AI-drafted text. Scores any document 0-100 against 43 rules on four axes: plain words, human voice, structure, information density. One JSON rule pack compiles to 18 targets - Claude Code skill, AGENTS.md, Cursor rule, system prompt, Vale style. Zero dependencies, runs offline.
Every AI session ends — usually with a wasted paragraph or silence. /wrap closes it with the outcome instead: tell the model how its work turned out, get one emoji back, and the record carries into the next session. Visible in memory, never a training reward. Open spec + prompts for every major tool. MIT.
Meta-skill toolkit: scaffold production-grade Agent Skills with versioning, baselines, packaging & release audit — initialize or adopt any skill repo in one command
Portfolio risk an AI agent can drive and a person can read. Historical VaR and Expected Shortfall, signed linear exposure, explicit stress scenarios, counterparty exposure and XVA through a local ORE project. 4 MCP tools, 5 skills, a local dashboard and a hosted service. Research software, not investment advice. Noncommercial licence.
Claude Code skill for removing backgrounds from AI-generated images using two-pass difference matting (white + black background pairs → transparent PNG)
A Claude Code skill for schematic & netlist review and hardware design decisions — with a deep-dive track for precision analog front-ends (TIA microcurrent, ΣΔ ADC, guarding, isolation).