alexgreensh

by Alex Greenshpun

[BETA] Token Optimizer diagnostic probe: proves which hook events fire in a Claude Cowork session (env dump + marker files + optional phone-home).

by Alex Greenshpun

Audit, monitor, and reduce Codex context waste with continuity helpers and explicit outline tools.

by Alex Greenshpun

The Token Optimizer status bar on its own, for its tests. It ships inside the token-optimizer plugin.

Audit, monitor, and reduce Codex context waste with continuity helpers and explicit outline tools.

by Alex Greenshpun

Token Optimizer for Claude Cowork: self-contained plugin with full skills and Cowork-native hooks (UserPromptSubmit, PreToolUse, PostToolUse, Stop). Install THIS in Cowork; the standard token-optimizer entry is for desktop Claude Code.

by Alex Greenshpun

Output styles that make Claude answer-first, plain, and easy to skim, plus a /tldr transform. Attention-kind, Spartan, and Rundown ship as both output styles and user-invoked skills.

by Alex Greenshpun

Code-drawn stills, drawing timelapses, films, explainers and interactive web animations in 31 styles, with scores composed in code.

by Alex Greenshpun

Code-drawn stills, drawing timelapses, films, explainers and interactive web animations in 31 styles, with scores composed in code.

by Alex Greenshpun

Audit untrusted repos before they touch your agent. Fully local, zero dependencies, zero telemetry.

by Alex Greenshpun

Audit untrusted repos before they touch your agent. Fully local, zero dependencies, zero telemetry.

by Alex Greenshpun

Make your AI coding tools work as one team. Route jobs across Claude Code, Codex, Cursor, Devin, Gemini, OpenRouter, and local models; carry the user's setup into delegated work; track limits and cash cost; and convene multi-model reviews.

by Alex Greenshpun

Decide, design, gate, judge, and defend evaluations for LLM, agent, and retrieval systems. Not ordinary unit tests.

by Alex Greenshpun

Answer questions from the real history of the 10x Marketers community. A verified member enrolls once (Google sign-in + community phone number), then their own agent asks questions conversationally; the skill retrieves cited community message chunks (multi-query, Hebrew+English, wide top-k, time-aware via since) as untrusted quoted data, and the member's agent runs trend sweeps and tool comparisons, surfaces evolution and contradiction rather than fake consensus, and says plainly when the corpus is thin. The skill retrieves; it never generates the answer.