Editorial Recension
A skill for text editing that deploys dual subagents subsequently to score and edit text both chunked and holistically, enforcing coherence and cohesiveness across editing passes. The scoring agent addresses implied knowledge gaps or assumes experience in a given field while the editing agent is passed the failing sections, their scores, and controlled context. They will keep working until each agent has been deployed five times or scores are not improving within three rounds and the user is notified. Best for non-fiction writers who value audience comprehension.
GithubGregorian Mode
A design-discipline plugin enforcing reasoning before decisioning in targeting two chronic failure patterns: reasoning-execution-collapse and conventional pattern gravity without contextual reasoning. Best for front-end builders, designers, and tinkerers who value deliberate design systems. Does not work well for quick mockups or vibe coding without creative engagement in the design.
GithubThe Docket Under Construction
The Docket is a tool that stops an AI coding assistant from quietly undoing decisions you already made about your project. It keeps a numbered list of your project's decisions in a file in your codebase, shows the assistant the relevant ones before it makes changes, and then has a second AI check its work when it says it's finished. It blocks it if there's a broken rule and tells the model exactly what to fix. Built for solo developers or small teams who keep watching their AI assistant forget why something was built a certain way and then "fix" it back into a new problem.
GithubWith love, math
A decision framework for agents built to loop agentic design decisions when building front-end work. This plugin aims to reinforce reasoning where agents typically collapse reasoning into conventional pattern gravity. Best for agentic design work where intentional design systems matter.
GithubAgent Plugins
Intention-driven Development
To keep this simple, I wanted to build tools for myself with AI but kept finding that the agent would make design decisions within its reasoning that had nothing to do with the context of what we were building - vibe coding I guess. But it robs the user of any creative agency in the resulting work. Something as simple as a color palette was impossible to maintain, so I built these 4 plugins to reclaim that agency and let AI handle the part its good at - strictly scoped coding.
With Love, Math sits at the top as the initial decisioning layer. It helps turn a user's vague design intention into a reasoned verdict by running it through a decision framework and providing measurable decisions with enforceable justification.
Gregorian Mode sits beside it as the design conscience. It prevents the agent from collapsing its own perfectly reasonable analysis into default design decisions (cards, tabs, buttons, layouts, etc.) by making every choice justify itself in both form and function at multiple scales.
Editorial Recension sits downstream of both. It takes the prose that comes out (the PRD, the UIUX notes, the ruling text) and edits it down to the knowledge level of the audience. Reasoning gets translated from an in-session model with a context window to a human-readable format that survives without context.
The Docket sits at the bottom as the final enforcement layer. This plugin takes those decisions and reasonings, writes them as numbered rulings in a DECISIONS.md file, cites them from the codebase, and enforces them as a stop hook. It refuses to let a session end if the work contradicts a ruling it already made.
These plugins exist as isolated features, but work coherently with each other because each one solves a different failure mode of the same underlying problem: an agent forgets, defaults, rushes, and drifts - even when spec driven documentation exists. The philosophy tying them together is that taste, intention, and governance are not separate issues.
Because you can't outsource creativity.
GitHub