The full vibe-coding loop: scaffold an app or marketing site with a real design system, animate it with GSAP and a real motion system, replace the fake data with a real backend, unstick it when it breaks, strip the AI look, ship it, and demo it.
★not rated 3 1mo agoA
tokens not measured
originalMIT
Strict multi-agent Supervisor for Claude Code: web-grounded anti-hallucination, expert coding/frontend implementation, independent judges, adversarial review, and final Stop release gate.
Turns a rough, long-running request into a loop-engineered contract and then executes it — frozen scope counted from the source, an achievement oracle written red before the work, a distinct proof per item, and a completion gate that runs the oracle itself so "done" is a fact rather than a claim.
★not rated 3 1mo agoA
tokens not measured
originalMIT
Turns a project repo into a narrated, edited demo video. Reads the codebase, writes a storyboard with real narrative value, prepares deterministic demo state, drives the UI with Playwright (web, launched Electron, or an attached running app) recording one clip per section with Playwright video or OBS over…
Chief of Staff pessoal pra quem trabalha com produto: um segundo cérebro de trabalho que nasce de uma conversa, mora em arquivos markdown seus e se molda a você. Rode /pm-chief-of-staff:setup numa pasta vazia.
★not rated 3▲
+1 1mo agoA
tokens not measured
originalMIT
Halftone paper-collage stop-motion ads: decode a reference into an editable spec, generate cut-out elements on the Gemini image API, and assemble build-from-empty motion with a deterministic local compositor, complete with the handmade stop-motion wobble. No paid video model.
★not rated 3 2mo agoA
tokens not measured
originalMIT
A memory that grows. Your agent is brilliant within a session and amnesic between two — this gives it a persistent knowledge trunk on your machine: what you work out once is distilled into a note, filed, linked, and surfaced automatically next time, from any project. It makes no request of its own — no telemetry, no…
★not rated 3▲
+1 23d agoA
tokens not measured
originalApache-2.0
Turn review into an organizational judgment asset. Review Judgment as Code for AI-assisted development, delivered as a capability pack for Claude Code: your team's review criteria as versioned, repo-owned skills (a Skill Registry), driven by a review agent that runs perspective-based reviewer roles (code, security…
★not rated 3
changed todayA
tokens not measured
originalMIT
Claude Code plugin for OpenAI Codex subagents, Codex Spark, parallel code review, adversarial review, and native follow-up sessions through a daemonless MCP server with read-only defaults and explicit full-access mode.
★not rated 3 2mo agoA
tokens not measured
originalMIT
Every Claude Code session, in every project, captured into one plain-Markdown brain you own. Index-first recall, automatic redaction, LLM-distilled session notes.
★not rated 3 1mo agoA
tokens not measured
originalMIT
Chop AI writing tells out of text. Wraps the slop-chop CLI so Claude can strip em-dashes, semicolons, and buzzwords, enforce a spelling dialect, or run the model rewrite pass on a draft.
★not rated 3 yesterdayA
tokens not measured
originalMIT
Do design work with Agentic Atlas over hosted MCP at agentic-atlas.dev/mcp/. Three skills — audit an existing agent artifact, design a new agent system, apply one named pattern — over two read-only agents; no SDK, CLI, local server, or content bundle is installed.
The human-in-the-loop for AI coding: verify every turn against the real diff, then make the agent run/screenshot/test and self-fix before finishing.
★not rated 3▲
+1 1mo agoA
tokens not measured
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: