cass-memory-system
Procedural memory for AI coding agents that turns scattered agent sessions into persistent memory shared across agents, so each agent can learn from the others' past work.
4,248 plugins found in public marketplaces and GitHub repositories, indexed and searchable. Plugin skills also install in Codex, Cursor and 4 more agents. In Claude Code, the 99 that get Auto-invokedWhat is this?This plugin ships a FLOW.md router the engine fires, so the matching skill runs itself. No slash command to remember.Learn how → fire as you prompt.
Procedural memory for AI coding agents that turns scattered agent sessions into persistent memory shared across agents, so each agent can learn from the others' past work.
A persona skill that answers programming questions in the voice and reasoning of creator Yupi (程序员鱼皮), covering learning paths, job interviews, resume tips, tech choices and AI coding. Works with Claude Code, Cursor and OpenClaw.
Generates type-safe SDKs, Terraform providers, MCP servers, CLIs and contract tests from an OpenAPI spec. Useful for API teams that want to ship client code and tooling for their users.
A Claude Code plugin with skills to search for jobs matching your resume, tailor a cover letter to a posting, fill out applications on Greenhouse, Lever, or Workday, and scan LinkedIn contacts' companies for openings.
Push it to GitHub and send us the link. We read it, then it goes live.
Lets your AI assistant send messages, pull data and trigger workflows across connected apps, using plain English instead of code. Guided onboarding walks you through setup and suggests actions for your role.
A design engineering system for AI coding agents that guides UI work away from generic AI-looking interfaces, with skills for dense dashboards, editorial layouts and minimal designs. Installs as a skill or a CLI.
The director remembers how your team works, and makes every agent session work that way. Your decisions, conventions and hard-won methods become a small memory of linked rules.
A set of Claude Code skills for mathematical modeling competitions, covering data harvesting and cleaning, model selection, modeling code and result generation, paper checklists, and task memory. It helps move from problem parsing to a finished competition paper.
Gives AI coding agents working knowledge of Unreal Engine C++ across actors, AI, animation, threading and audio, so generated code uses current APIs and avoids outdated ones.
A writing workflow plugin for Chinese long-form and WeChat articles that turns drafting into a collaborative process with planning, editorial review, revision, and fact-checking. It includes skills for style modeling and web article extraction.
Skills for generating Java project inventory checklists of commands, agents, and skills, and for installing embedded commands and agents into Claude, Cursor, and Codex directories.
Checks an AI agent's tool calls before they run, blocking unsafe actions and scoring output against declared rules. Rules can look at what already happened, so a policy such as checking before a refund becomes one rule instead of prompt text.
Reviews a UI design and gives targeted fixes for hierarchy, typography, color, spacing and buttons, so you can tighten a screen without guessing at each decision. Run one fix on its own or a full design pass across them.
Helps you write AI video prompts that work, picks the right model for your clip, and translates or debugs prompts across Kling, Seedance, HappyHorse and other video models, so you stop trial-and-error regenerating.
A GEO (Generative Engine Optimization) toolkit of skills and commands for auditing content quality and E-E-A-T, writing SEO-optimized pages, and building client review pages. Open source under MIT.
Lets Claude Code control a Chrome browser session through the chrome-devtools-axi CLI, to navigate pages, read accessibility snapshots, click, fill forms, run JavaScript, inspect console and network activity, take screenshots, and audit performance.
Lets your AI coding assistant handle everyday tool work for you, such as tracking Jira and Confluence items, managing Azure DevOps pipelines, and turning text into narrated audio, so you stop switching between apps.
Runs shell commands and AI agents inside a limited user account on your Mac, and can clone the current repo into a sandboxed Claude session for a task.
Gives a coding agent a repeatable path from a clear change request to a checked result. It implements fixes without weakening tests, can deliver an independent go or no-go verdict on finished work, and keeps project memory so the next session starts with context.
Runs image cropping and rotation and fillable PDF form text replacement inside a sandboxed container on your own machine, so sensitive files never leave it. Works with any MCP-compatible tool.
Gives your coding agent a list of tasks or an open goal and a time budget, then keeps it working until the list is done, leaving decisions parked for your review. Progress survives across sessions, so you come back to finished, reviewable work.
Gives journalists and researchers ready-made guidance for fact-checking, source verification, interview prep, story pitches and data journalism, plus accessibility audits for news sites. Works in Claude Code and Codex.
Rewrites the previous assistant message, or any text typed after the command, in much simpler plain language. Run /bro when an answer is hard to follow.
Generates ready-to-import ComfyUI workflow JSON files from plain language descriptions of the image you want. The output is valid API-format JSON that you load into ComfyUI and run.
A package manager for AI agent skills, letting teams install, share, sync, and create them across coding tools such as Claude Code, Cursor, and Codex. Includes a debrief skill to fold session feedback back into a skill.
Gives your AI coding assistant a memory across sessions: it saves task threads, searches past context semantically, restores where you left off, and versions changes with Git. It also adds specialist roles for backend, frontend, and DevOps work.
Gives your coding agent skills to author, maintain, validate and visualize Open Knowledge Format bundles, portable markdown knowledge that both your team and your agents can read. Includes a backfill skill for building a bundle from an existing repository's git history and session transcripts.
Lets an agent operate Scrumboy project boards through MCP or Agoragentic APIs: inspect board state, plan sprint work, create todos, move cards, manage tags, and summarize project health.
Generates e-commerce product images and campaign photo sets with GPT-Image models through a user-configured OpenAI-compatible endpoint, using scene templates. It can also export prompt packs for manual use.
Lets you dictate into any macOS app with a hotkey. Speak, and the corrected text is pasted where your cursor is, with a personal dictionary to fix recurring words.
Drives Smithers durable flows that run ordered agent stages with retries, approvals, bounded loops, and crash-safe work. Use it to run existing flows or author new TypeScript flows with Flow.make, Action.make, and Effect.
Lets an AI agent drive XCrawl, a web data service, to search the web, fetch and extract pages, map a site's URLs and crawl whole sites. Each task gets its own guidance, so the agent calls the API consistently instead of you writing the requests by hand.
Connects plain-English requests to specialist agents, skills, and workflows for Claude Code, covering code review, code quality, assessment, data analysis, and business operations. The /do command picks the tools a task needs.
My agent skills that I use every day to do real engineering - not vibe coding. Developing real applications is hard. Approaches like GSD, BMAD, and Spec-Kit try to help by owning the process.
Solo-business skills for Claude Code covering advisory meeting minutes pushed to Notion, book-style PDF and EPUB export with cover and running headers, community launch planning with Notion setup, Claude-to-Codex fallback on usage limits, and desktop app control.
A local-first personal knowledge base plugin for Claude Code. It builds and updates a Markdown wiki from your files, extracts WeChat articles and converts documents, PDFs and media to Markdown, and renders videos from notes.
A skill that reviews Apple platform app code for App Store rejection risks before submission, checking it against Apple's App Store Review Guidelines for iOS, macOS, tvOS, watchOS, and visionOS.
Catches accessibility problems while an AI coding assistant writes web interface code, so WCAG accessibility issues are caught before they ship. It also compares audits across commits to flag regressions and drafts an accessibility statement from the findings.
A Chinese-language bidding assistant for Claude Code that interprets tender documents, generates bid text, checks bids for compliance, and formats the final Word document.
A healthcare plugin with skills for extracting structured data from clinical notes, drafting clinical trial protocols, answering questions across contract corpora, converting documents to text, and building FHIR API endpoints. It also connects hosted MCP servers.
Compiles your Claude Code conversation logs into compact views, so a long session that was compacted or ran out of context can be recovered in its original detail. Also lets you search and reread past conversations without digging through raw JSONL files.
Skills for building, debugging, and configuring Output SDK workflows, the open-source TypeScript framework for AI workflows and agents. Covers implementing workflows from a plan, debugging failing runs, and managing encrypted credentials and environment variables.
Gives AI agents step-by-step instructions to control Android and iOS devices, locally or in the cloud, handling taps, swipes, screenshots and app control for you. Works through ADB, Portal or the simulator with a Python control library.
Workflows for knowledge work that build on past sessions. Brainstorm, plan, check confidence, review drafts for strategic alignment and data accuracy, and save learnings to docs/knowledge/ so future plans find them.
A Claude Code skill providing the latest OWASP security best practices (2025-2026) for developers building secure applications.
NotebookLM-style AI podcasts, except you control everything. The script. The prompt. The hosts' roles. Their voices. The content they discuss. All of it. A coding agent skill that turns any content into a two-host podcast episode.
Skills for job searching: analyzing job postings, building and tailoring resumes for ATS systems, and preparing for interviews including behavioral questions. Aimed at developers looking for their next role.
Scans source code, dependencies and repositories for security weaknesses such as injection flaws and leaked secrets, then merges the findings into one prioritized report. It also captures and replays HTTP traffic through a proxy to help test running applications.
Gives Claude Code persistent semantic memory across sessions, storing user preferences, project context, prior decisions and codebase facts. Claude can search this memory before answering questions about past work.
Helps coding agents solve software engineering problems by finding proven implementation paths in GitHub repositories, issues, and pull requests. Built for developers debugging dependencies and comparing approaches.
A terminal orchestrator that runs headless Claude Code agents through a gated delivery pipeline for any repository, covering planning, decisions, issue breakdown, test-driven building, and end-to-end checks.
Searches arXiv for prior art and related papers on your topic. Fires when you ask about existing work, eval design, or reasoning architectures before building.
Covers the full math modeling contest workflow, from solving a problem to writing the paper, for the China CUMCM and US MCM/ICM competitions. It pairs a solver skill for modeling approaches and code templates with a paper-writing skill for structure, chapters and formatting.
A Claude Code plugin of agents, skills and commands for go-to-market teams. It covers account tiering and ABM personalization, consolidating intent and usage signals, scoring account health, and drafting executive briefings.
Finds functions that do the same job under different names, queries MCP servers only when you ask so they stay out of your context, drives interactive tools like vim or a Python REPL through tmux, and sets up a headless Windows VM you can SSH into.
A local-first memory layer and MCP server that stores session history, codebase indexes and concept graphs in SQLite for long-term project recall. It runs without cloud storage and works with Claude Code and other MCP clients.
Lets AI agents read and update your monday.com boards and CRM records directly, so you stop copying data between tools. Ready-made skills cover morning briefings, sales forecasts, rep activity reports and data cleanup.
A local GUI for Agent Package Manager packages. Import agents, skills, prompts, hooks, and MCP configs, then sync them to Codex, Claude, Cursor, Gemini, OpenCode, and more.
Claude Code skills for copywriting, advertising, SEO, design, and more. Each skill is a self-contained markdown file that gives Claude deep domain knowledge. Drop one into your project and invoke it with a slash command.
A local dashboard for watching OpenClaw coding agents in real time, showing costs, tokens, sessions, tool calls, memory and system health. It also includes a kill switch to stop a runaway agent.
Behavioral guidelines that help Claude Code avoid common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable goals.
A file-backed workflow harness for Claude Code and Codex that covers repo setup, product requirement drafting, cross-review of changes, and project memory saved to Obsidian on request.
A Claude Design skill that turns UX decisions into a step-by-step blueprint animation. Show a redesign from Before to After with changed parts rebuilding, or annotate a single screen module by module with what and why, for case studies and posts.
Turns Xiaohongshu posts into Obsidian notes, with transcription and image OCR. Fires when you save a Xiaohongshu link, batch-import posts, or ask for a cover or analysis.
Lets an AI assistant work with your Odoo database through guarded tools, so agencies can review many client databases, check data quality before trusting answers, plan version upgrades, and run month-end accounting closes with human sign-off at each step.
An intent-driven workflow runtime for AI coding agents that keeps context, skills, approvals, validation, and handoffs consistent across sessions and hosts. It bundles skills, agents, and MCP servers for orchestrating complex coding tasks.
Automatically creates, enhances and maintains CLAUDE.md files for Claude Code projects, so you don't have to write and update them by hand.
Claude Code Install: Update: Auto-update: off by default. Turn it on in /plugin → Marketplaces → dryforge → Enable auto-update. Codex Install: Update: Auto-update: on. New releases load at the start of each session. Grok Build Install: Update: Auto-update: on.
A structured skill framework for Claude Code covering deployment, security audits, content pipelines, scheduled jobs, API integration, and workflow automation. Skills load on demand when a matching request comes up.
Claude Code Install: Update: Auto-update: off by default. Turn it on in /plugin → Marketplaces → dryforge → Enable auto-update. Codex Install: Update: Auto-update: on. New releases load at the start of each session. Grok Build Install: Update: Auto-update: on.
Turns raw evidence from cloud accounts, SaaS tools and code into framework-aligned findings, gap reports and remediation steps, so compliance checks become repeatable and easier to audit.
Turns an article idea into an original print-style editorial illustration, with a recurring mascot acting out the point. It can also build labeled explainer diagrams such as flowcharts, timelines and fan-outs, so you skip hand-drawing visuals for your writing.
Gives Claude Code skills to work inside a live marimo notebook: run Python in the same kernel the user uses, inspect live notebook state, and commit notebook changes. Includes a skill for reviewing pairing sessions and improving the tooling.
Gives coding agents focused skills for turning a rough idea into reviewed work: checking a technical proposal before coding, designing the system architecture, and delivering tested pull requests through review and gated merges.
Vigil is an open source, agentic AI SOC released under Apache 2.0. Your playbooks are plain-text Markdown files, your agent logic is readable code, and your integrations use an open standard (MCP). It is a capability you own, not a black box you rent.
Plugin for the Station CLI (stn) to create, run, and deploy AI agent teams, with skills for configuration, code review, and web research. Self-hosted and git-backed.
Helps Claude Code users avoid burning through their usage quota by automatically rotating oversized sessions and carrying context across them. Includes a skill for saving a finished workflow as a reusable skill.
A multi-agent orchestration plugin for Claude Code that runs design review gates after brainstorming, turns issues into autonomous PR workflows, handles PR review comments, and delegates tasks to Codex and Gemini CLI tools.
A harness for coordinating teams of AI agents using SAFe-style workflows, with skills for agent assignment, API route patterns, documentation templates, and deployment checks. Built for software teams and adaptable to other repeatable processes.
Searches, saves, and maintains a local document index through a local RAG MCP server, so private documents stay on your machine with offline embeddings instead of being sent to an embedding API.
Security rules and a skill that help AI coding agents write and review code with secure-by-default practices. Includes validators to test whether generated code follows the rules.
Prunes bloat from long Claude Code sessions, removing stale reads, duplicate content and progress ticks while keeping what matters. Includes a background guard that trims sessions before context overflow and helps protect Agent Teams from context loss.
round_pushpin: My dotfiles for macOS using Neovim, Zsh, and Ghostty + Tmux.
The ultimate Roblox game development Claude Code skill. A self-contained, full-lifecycle development companion that serves all skill levels with deep Luau expertise, autonomous game building via MCP Studio integration, and comprehensive knowledge of game.
Skills that build and maintain a Markdown guide to a repository, helping humans understand its real behavior, concepts and evidence. A Chinese-language variant keeps source identifiers for lookup.
A skill that lets Claude Code and other Agent Skills clients call the Agnes AI generation APIs for text, image, and video. Covers text-to-image, image-to-video, and keyframe animation.
A skill guide for the FinLab quantitative trading package, covering backtesting and strategy work across stock and ETF markets in Taiwan, the US, Korea, Japan and Hong Kong. Aimed at investors and analysts building trading strategies.
A command line tool for HEY that lets you read and send email, manage contacts, labels, calendars, todos, habits, time tracking and journal entries from the terminal or from your coding agent.
Switches Claude's replies to terse, caveman-style prose to cut output tokens while keeping technical accuracy. Offers adjustable intensity levels, from light to heavy.
A Claude Code skill that makes a showreel-grade product film in code: a landing-page loop, a launch video, a promo or a demo reel. It uses Remotion, your product's real components, its design tokens, its logo and its voice.
Coordinates multiple Claude Code agents on a task using a Royal Navy squadron structure, covering mission planning, parallel work and handoffs, then stand-down. Suited to work that needs several agents running in parallel.
Runs Warden code review on local changes before commit, sweeps a whole repository for issues and drafts PRs for validated findings, and queries the Warden Service read API for runs and findings.
Detects AI-writing patterns in prose files your agent saves, and includes a skill that helps humanize text and remove AI-sounding phrasing. Runs locally using only the Python standard library, with no network access or model.
Lets an AI agent operate a phone like a person would, watching the screen with a camera and tapping it with a stylus. Includes skills for shopping in the JD and Taobao apps.
A coding agent skill for creating polished, professional Reveal.js HTML presentations using natural language.
Your AI Desktop Office Assistant, Just Leave It to Abu A locally-run AI desktop assistant inspired by Claude Code's Cowork mode. Tell Abu what you need, it reads files, runs commands, writes docs, and builds reports, all on your machine.
Audits decompiled WeChat mini-programs for hardcoded secrets, exposed endpoints, and weak crypto. Runs a full static security analysis from decompilation to report.
Platform-neutral guidance for installing, verifying, troubleshooting, and operating Open Browser Use, an open-source Chrome automation stack for AI agents.
A portable skill that converts long-form Markdown, such as specs, RFCs, runbooks and postmortems, into a single self-contained HTML page with Mermaid diagrams, step timelines, callouts and a sidebar table of contents.
Create, animate, and export pixel art sprites in Aseprite using natural language in Claude Code. Covers canvas creation, frame-based sprite animation, dithering and palette work, and export to PNG, GIF, or spritesheets with JSON metadata for game engines.
Runs registered ComfyUI workflows through a command-line tool and pairs with the local Comfy MCP for templates, nodes, models, and validation. Aimed at AI agents such as Claude Code and Codex.
One Java engine for web, mobile, API, CLI, database, and desktop. Drivers, waits, and evidence are handled. Every run ends in an Allure report.
A traditional feng shui advisor skill covering Xuan Kong flying stars, Ba Zhai, date selection and home layout, for analysing living spaces, picking auspicious dates and resolving negative energy. Works in Claude Code and other Agent Skills compatible tools.
Generates new raster images and looping GIF or WebP animations from Claude Code using the local image-use CLI, with your existing ChatGPT subscription and no API key. It is a single stdlib-only Python file.
Gives your AI assistant a persistent memory tree, so what you learn in one conversation carries into the next instead of resetting. It connects new topics to your earlier knowledge and saves useful findings as you go.
Generates raster images and looping GIF or WebP animations from a local one-file Python CLI, using the user's existing ChatGPT subscription instead of an OpenAI API key. Usable by Claude Code as a skill.
Gives Claude Code and Codex a repeatable development process for planning, reviewing, and shipping changes, so you stop rebuilding the same prompts each time. It also captures corrections and project conventions so future sessions build on what you have already learned.
A plugin of skills for multi-agent review and debugging: adversarial challenge of ideas, structured code review, Lead-driven Hydra workflows, root-cause debugging, and browser-based QA testing.
Code and examples from my YouTube channel tutorials.
Runs test-driven development with a coder and overseer subagent that gates each checkpoint. Fires when you write plans, execute them, debug, review, or iterate on code.
A Claude Code plugin for offensive security work, offering structured red-team engagement workflows based on the Cyber Kill Chain. It covers attacks on Active Directory, AI and LLM systems, command and control infrastructure, and browser exploitation.
Open-source control plane for your AI agents. Connect tools, hire agents, track every token and dollar.
Agent skills for end-to-end fiction writing in markdown, covering chapter drafting, character tracking, discovery drafting, audiobook adaptation and editorial review. Packaged for Claude Code and Codex.
Generates AI images from text prompts through the nano-banana CLI, using Gemini Flash by default with Pro available. Supports several resolutions, aspect ratios, and reference images for style transfer.
Helps an architecture studio set up its own AI practice, covering how AI is governed in the studio, which practice areas it covers, and what it remembers. It gives architects ready-made skills for tasks like color palettes, FF&E schedule exports and site demographics.
A skill that applies a restrained Scandinavian visual system to a codebase, with a black-and-white foundation, sans-serif type, generous spacing, and purposeful product imagery. It is built for Cursor agents.
A Claude Code skill that sends tasks to background AI worker agents using checklist-based plans, so you can delegate work such as reviews to other models while you keep working.
© 2026 Flowy
Built for Claude Code · Not affiliated with Anthropic
Logos are trademarks of their owners.