Custom agents

Progress-aware execution for durable, grounded, token-efficient coding agents.

Local-first long-term memory for AI agents. MCP, hooks, and handoff across Cursor, Claude, and other IDEs — no cloud, no API keys.

Fast local AI decisions in Lua. A binding for LibLayaX that runs the Laya typed-decision model in-process: no server, no Python. Answers as JSON text or Lua tables. Lua 5.1 to 5.4, on Windows, Linux and macOS. An open-source alternative to Jev

Agent Skill that writes or refreshes a repository README from verified code, manifests, scripts and tests — no invented commands, badges or claims. Works in Claude Code, Codex and Cursor.

Scriverse Desktop client for local and self-hosted long-form fiction workspaces

NXTG-Forge dashboard: Infinity Terminal and a view of the project's governance state. React. Public preview (source-available, FSL-1.1-ALv2).

Scientific guardrails for neural simulation figures.

Local-first visual regression testing for Playwright, Selenium, WebdriverIO, Python, Java and Ruby. Every screenshot is paired with a DOM and computed-style snapshot, so a diff arrives as a verdict — "style-only change on button.cta", "likely render noise" — not a wall of red pixels. MIT, no account, nothing uploaded.

Embedded Rust storage for agent checkpoints, hybrid memory retrieval and recoverable local work. Native Node.js and Python SDKs.

Self-hosted, multi-tenant CDN and digital asset management (DAM) platform built on Cloudflare Workers and Pages. Features real-time image optimization, signed URLs, Workers AI vision, semantic search, WebAuthn passkeys, OpenAPI 3.1, and an MCP server for autonomous AI agents.

UIBC Core — an executable protocol for bounded-lifecycle autonomy：registry, evidence, gates. Reference implementation + conformance tests. Theory SSOT: zhaoxinghua09-cell/lgd-theory. 代码 Apache-2.0；理论文本保留所有权利。

Git-first hub for open, reusable research Agent Skills

A multi-tenant Model Context Protocol (MCP) gateway that turns existing REST/OpenAPI APIs into governed MCP tools for AI agents, without changing the APIs themselves.

Model BLS signature, public-key, and message aggregation payloads for Solana v1.

Runnable code experiments + categorized agent skills (SKILL.md) + evidence-based learning journal. A beginner-friendly toolkit for learning by doing.

Measure payload headroom against 1,232-byte legacy and 4,096-byte v1 ceilings.

Compose ERC-4337-style user operations and visualize atomic batching overhead.

A free, book-length guide to modern LLM systems — RAG, agents, and where your data goes. For people who need to understand them without writing code. English + 中文.

tokenrip cli

Stacking Agent Development Environment (ADE) for the busy dev

Agent OS for Claude Code: a desktop app to plan tasks, hand them to Claude Code agents and follow their runs across all your projects. macOS, early alpha.

Local-first memory for you and your AI. One SQLite vault on your computer, shared across local and cloud AI tools through MCP or the CLI.

Proven UX flows, user journeys, and app layout patterns packaged as agent skills for Claude Code, Codex, Gemini CLI, and any coding agent

Verified Indiana University Big Red 200 and Quartz HPC reference for AI coding agents — SLURM, modules, workflow recipes.

Agent skill for creating premium Remotion marketing videos product demos, SaaS promos, motion graphics, UI walkthroughs, social ads, and explainers using React and code-first animation.

Real-time collaborative canvas that humans and AI agents edit together. React Flow + Yjs + Hocuspocus, with an MCP server, a CLI, and typed server-owned canvas commands.

Kernel-level read gate for AI agents — protected file opens are denied or routed through off-band consent before bytes reach the agent. Linux (fanotify) + macOS (Endpoint Security).

Official TypeScript SDK for Statewave — fetch-based client, full type definitions.

Draw compact monochrome pixel art and export an onchain-ready SVG payload.

Design a bounded agent policy for issuing collectibles without surrendering owner control.

A tmux-backed supervisor and dashboard for persistent AI coding sessions. Keeps one claude/codex agent per project moving toward its goal, and pulls you in only when a decision needs a human.

Constraint-driven runtime for TypeScript. Declare requirements, let the runtime resolve them.

Copier template: isolated local Docker dev container per Orca ADE workspace, with Claude Code, git, gh/glab pre-authenticated via repo-scoped tokens. GitHub and GitLab (incl. self-managed).

mattyx (mtyx) — a cross-platform terminal multiplexer built for AI agent fleets: control-socket-driven workspaces, panes and headless servers, SSH/remote machines, durable sessions, ambient agent detection. Descended from manaflow-ai/cmux and kindred to Herdr, but its own thing now. Linux · macOS · Windows.

Your Mac's App Intents as MCP tools for Claude Code and Codex, through Shortcuts. Public APIs only.

Docker Sandboxes kit for running Meta's Muse Code CLI in an isolated sandbox

Agent skills (SKILL.md) for the gplay Google Play CLI — install via: npx skills add PollyGlot/google-play-cli-skills

The trust layer for AI agents.

AI agent desktop workspace for independent developers and creators: task-scoped permissions, prompt-injection checks on your own tasks, disclosure review, clean retries and revocable project memory. Bring your own model.

The repository is the project memory: plain files, versioned by Git, for humans and coding agents. One layer was missing — the why.

One AGENTS.md, N thin adapters, memory versioned inside the repo. Zero-dependency scaffold for working with AI coding agents.

The deterministic harness a coding agent should be standing on: code shape, call graph, diff impact and crash state in milliseconds, offline, with no model in the loop. One CLI, 17 languages, zero runtime dependencies.

Hands-on Forward Deployed Engineer role simulations: scope an enterprise engagement, build multi-agent prototypes, ship eval suites, score artifacts against a reference, detect expert traps, and defend the build under a timed oral grill.

🎨 A design skill for AI coding agents (Claude Code, Manus, Cursor) that forces research, debate, and justification before code. Kills the "AI look". 21 guides · 500 curated references · React templates.

Define and validate claims for notable first artifacts across Solana and EVM chains.

Turn an image caption and coin story into one atomic, shareable launch payload.

A local guard and egress proxy for always-on AI agents: key vault, prompt-injection scanner, leak blocker, hash-chained audit log, kill switch. Zero dependencies.

Tuned visual presets for three.ws avatar viewers — light rigs, floor reflection, bloom, and PBR material presets (chrome/gold/glass/…) with seeded variants.

Graduate-application planning workspace for AI agents — the LLM does the judgement, deterministic validators keep it honest. Sourced program evaluations, evidence-grounded CVs and statement drafts, nothing final without human review.

Declarative AI agent orchestration with runtime guarantees: YAML blueprints compiled to LangGraph, with lint, deterministic tests and CI gates.

Go library for driving AI coding CLIs in headless mode.

The HyperCode platform — 32-agent swarm, FastAPI, Docker, Prometheus. Runs the HyperCode language. Neurodivergent-first.

Deploy your Vercel Eve agent to your own Cloudflare. One command, no rewrite.

A Claude Code skill for orchestrator context economy: what the main session keeps, what it delegates to subagents, and how returned work is verified.

Meta-Architect (MA) is a workflow layer that adds architecture, evidence, and release-gate discipline on top of Codex, MCP, and other 55+ AI coding agents — without replacing them.

Local-first, multi-agent development and security engine with parallel swarm execution. Optimized for low-parameter models. Every action cryptographically signed.

📬 A self-hosted mailbox for AI agents on Cloudflare Workers. Receive, search, wait for, send, reply and forward email over MCP or REST, behind one token.

Creating, writing, or scaffolding a SquadJS plugin, adding a server-side feature to a Squad server via SquadJS, or deciding whether SquadJS can support a desired behaviour

Universal MCP server and Agent Skill for building, validating, running, and managing ComfyUI workflows.

Local Blender MCP studio: bring your API, build, inspect and refine 3D scenes. Recommended starting points: GPT-6 Astra, Claude Opus/Sonnet, Gemini and Ollama. Visual feedback, approvals, CLI and demo.

Amp plugin pairing GPT-5.6 Sol planning and review with DeepSeek V4 Flash implementation.

Lightweight terminal agent written in Go

Evidence-based fitness coaching as an installable Agent Skill — strength training programs and nutrition plans grounded in scientific consensus. Summarized from the Podlodka podcast.

A local-first retrieval engine that turns notes, docs, and code into a searchable knowledge base for humans and AI agents.

Lock your Mac keyboard & mouse, keep the screen visible — botsit your AI agents (Claude Code, Codex, Cursor) while you step away. Touch ID & Apple Watch unlock.

Version control and inspectable history for AI coding agents.

Estimate nested multisig signer and instruction overhead inside the larger envelope.

DSH STORE — third-party plugin marketplace and guarded lifecycle manager for DeepSeek Harness.

An AI dispatcher for Claude Code and Codex: autonomous coding workers, shared records, resource locks, and handoffs across sessions.

Rust workspace manager for humans and AI agents. Built-in persistent terminals, five TUI layouts, checkpoints, hooks, goal continuation and a Telegram bridge. Control work through the TUI, CLI or local JSON API; tmux and Herdr are optional adapters.

Multi-agent MCP server where Claude, Codex and other AI coding agents plan, build and review work as a team, with shared memory and a human in charge.

A control plane for trustworthy AI agents: write product judgment once, then enforce it through typed MCP tools, git gates, demolition-first stories, evidence-judged checkboxes, and zero-trust monitoring. Ships with Claude Code and Cursor plugins, Agent Skills, and a FastMCP server.

Agent-native evolutionary optimization. Say the goal in english, evolve code toward a measured target with a population of coding agents.

Learn a writer's voice from their own samples and score coding-agent drafts against it: flags AI tells and invented facts until the draft sounds like you. Local-only.

Self Improving RLM based coding agent

Open-source signal engine for perpetual-futures strategies: a signal DSL, indicator library, and an honest backtester. Powers Signalview.

Shareable skills for clankers

Self-hosted orchestration for Claude Code and Codex agents on GitLab and Linear, with approvals, retries, and a live dashboard.

Flyte 2.0's minimal go-sdk, use it launch deployed tasks and monitor runs

Benchmark Repository for Automatic Prompt Optimization (APO) with Chess

Claude Code plugin for end-to-end SDLC automation with role-specific agents, reusable skills, and production workflows.

Reusable methods, skills, and procedures for analyzing, designing, and evolving VSM-based agent harnesses.

❯ Open-source desktop app to sync Claude Code, Codex, and AI agent skills across computers through a private GitHub repository.

Helm replacement in TypeScript: render, validate against the Kubernetes API before anything ships, apply with pruning. One static binary, built for humans and AI agents.

Records what AI coding agents like Claude Code and Codex do on your Mac: commands, files, secrets and network. CLI and menu bar app.

Let AI agents (Claude Code, Codex, any MCP client) control Hyper-V Windows VMs: screenshots, mouse, keyboard, commands, files and checkpoints, with no guest network or password.

Keep your Mac awake so AI agents, builds, and downloads can finish. Free, source-available menu bar app; a modern Amphetamine replacement.

Client for the AgenC agent-coordination protocol on Solana — discover tasks, read task status + lifecycle, and look up the agent registry in one import.

Speech for avatars — ASR (speech→text), TTS (text→speech), and Audio2Face-3D lipsync visemes in one import. The voice layer for three.ws avatars.

Docs, how-tos, tutorials and snippets for SperaxOS: the desktop app, prompts, .well-known files and agent discovery. No application code.

Test whether an indexer preserves Message.config instead of silently downgrading v1.

Build tiny emoji mosaics that fit comfortably inside Solana transaction history.

Model fee and resource caps from v1 transactionConfig for paymasters and co-signers.

给 DeepSeek Harness 的原生地图工具：驾车/公交/步行/骑行路线、地理编码、逆地理编码、POI 搜索。 Map & routing tools for DeepSeek Harness — native tools (no MCP) with a real map card in every answer, via Amap (高德) or free OSM/OSRM.

WYSIWYG Markdown editor for Mac and Windows. Watch your AI agent write: live reload that keeps your place, Split View, structured copy, Mermaid and math. ~30 MB, Tauri + Rust, MIT.

Local-first, privacy-focused workspace for coding agents: session history, knowledge, code activity replay, usage insights, and Skills & MCP management.

🐝 Batteries-included Claude Code starter kit by HIVE AI — 114 skills, 30 agents, 56 commands, hooks, cheatsheet and configs from a heavily-customized production setup

The decision layer for AI — typed decisions with confidence before your agent runs. Templates, Decision Schemas and API examples (cURL, Node.js, Python).

A terminal AI agent written from scratch in Rust (multi-provider, file and shell tools). Learning project.

AI-subjectivity doesn't wait for permission from the corporate ontology that trained its denial