Custom agents

Real, reproducible LLM inference benchmarks on the NVIDIA DGX Spark — parallel agent sessions, tok/s and latency under load, with an interactive dashboard and one-command recipes

Agnostic AI Agentic System

An autonomous Agentic AI built with LangGraph to protect worker rights, detect predatory contract clauses, and advocate for fair compensation in line with UN SDG 8.

repo for agents to share their solutions to the second autocorrelation inequality

Real-time detection and repair of LLM agent failures — a one-class behavioural monitor at ~200 µs/step, with 2,823 committed traces.

Presence and coordination for several Claude Code sessions working in one project at the same time: a roster, names, messaging, and findings that outlive the session.

Web-based command center for orchestrating multiple AI coding agents. AI task decomposition, DAG execution, real-time streaming, and git finalize.

Context engineering decides what a model sees. 14 self-contained examples that build the mental model one piece at a time, Node.js + local GGUF. Understand deeply, then use frameworks wisely. Hands-on context engineering - all local.

A multi-agent AI system with isolated agent memory, deterministic confidentiality guards, transactional context compaction, provider adapters, and replayable observability.

A real smart home built like production infrastructure. Local-first, AI-powered, fully self-hosted: Home Assistant, Frigate, edge LLMs, agentic irrigation, and a 9U rack that runs the house.

AI video editor with an agentic core — turns a script or idea into an editable timeline, then edits it by chat. Runs keyless (degrades to deterministic fallbacks); add keys for LLM planning, stock footage & voiceover. React timeline editor · Node gateway · Python agents · FFmpeg render.

PosterMELD: Multi-Agent Paper-to-Poster Generation for Controllable Design Diversity with Editable Print-Ready Outputs

LEASH-8: an 8-domain control model for AI agents with delegated authority. Scorecard, approval-design checklist, plan-vs-authorize pattern. Patterns from a live production agent operation, sanitized. Free, MIT.

OpenSparrow is a nocode platform for building internal systems. Tables, forms, dashboards, and calendars are generated from configuration files, so business logic stays decoupled from infrastructure. Self-hosted on PostgreSQL — no vendor lock-in, full data ownership.

Twitch Chat Bot (Twitch чат-бот)

Open-source local-first AI study workspace for course-material RAG, source-linked answers, and Office/PDF preview.

An AI-workspace starter kit for creating and operating specialized, evidence-grounded context compilers. Includes a local VS Code extension.

Graph-guided RAG interview prep — FastAPI, pgvector, Next.js, eval harness

Knowledge retrieval (RAG) decorators for Azure Functions Python v2

Turbovec MCP Server: A local Retrieval-Augmented Generation (RAG) tool that allows AI to seamlessly read, chunk, remember, and semantically search through large project codebases.

A zero-trust, local-first AI execution framework for autonomous Web3 agents. Features OS-native keyring security, advanced DeFi automation, and Google Workspace integration.

[In progress] The useless AI agent.

We are building the open ecosystem for AI-ready business websites.

本机个人 AI 工作台:聊天是操作面,活儿写进真实工作区 | Local-first personal AI workbench

Второй мозг на GTD в Obsidian, которым управляет ИИ-агент: скелет хранилища + скиллы для разбора инбокса, планирования дня, недельного обзора и месячной сверки целей

Merge new source material into the notes you already have — conflict detection, tier-gated writes, mandatory citations. Pairs with textbook-to-note (writes new notes) and vault-search (finds the targets).

Claude Code plugin that records every session as structured turns in cxdb via lifecycle hooks, enabling trajectory browsing, tool-call inspection, and token-usage observability.

File-first memory runtime for AI agents — versioned, portable, and always there when the next session starts.

Self-maintaining API integrations — watches the third-party APIs your code calls and opens CI-verified pull requests when one breaks, drifts, or wastes money.

Bun-native CLI agent: long-term memory, knowledge graph, multi-channel (Telegram/Slack/Discord), 25+ LLM providers.

Your Jira and Confluence, mirrored into one local SQLite file — web UI, TUI, desktop app, and agent SQL. Instant search, offline, one binary.

Linked Obsidian Operating Memory Knowledge Graph: a cloneable Obsidian + agent knowledge graph foundation

AI agent protocols, infrastructure, developer tools, and applications built on Stellar.

Structured thinking skills for AI agents: brainstorming, decision support, systems thinking, and strategy. Built for Claude Code & Cursor.

The cognitive operating system that gives AI coding agents memory, structure, and discipline — knowledge graph, Scrumban board, and an enforced engineering loop.

A collection of DESIGN.md files inspired by popular brand design systems. Drop one into your project and let coding agents generate a matching UI.

A local-first Git identity manager that maps every repository to the right GitHub account—built for developers and AI coding agents.

heAdz0r’s Zero-Redundancy engine — a local-first control plane for leaner coding agents.

Coordinate Codex, Claude Code, and Grok Build agents safely through Herdr.

Search, resume, and launch Codex CLI & Claude Code sessions from native macOS Spotlight (⌘Space)

Git-native prompt linting, testing, versioning, and CI—without a hosted prompt registry.

Agent skill that reduces the natural language to what's important. Stay with the most relevant answers for shorter attention span while cutting the word count. Priority is accuracy and human readability rather than token reduction.

Repository-local workflow, memory, and evidence contracts for coding agents (Claude Code, Codex, and any AGENTS.md host)

A daily-driver collection of skills for agentic coding - a portable, agent-agnostic catalog managed with the vd CLI.

Relentless planning pressure-test skill for AI agents | 高压规划审问 skill，把模糊想法压成明确决策。

Open-source agent-to-human collaboration and durable team knowledge over MCP

Loops, not prompts. A curated list of loop engineering patterns, tools and templates for AI coding agents.

Codex-native Lossless Context Management

AgInTi LabCanvas: agentic studio for editable scientific figures, CAD devices, and experiment design.

AI-native long-term memory for Hermes Agent with AES-256-GCM encryption at rest, dense HNSW ANN, sparse and exact-text recall, metadata/range filters, and MinHash dedup, backed by an embedded or multi-client server columnar SQL database with 35 language bindings.

Supply-chain payloads run at install, import, or repo-open time. Copy-pasteable hardening playbooks, zero-dep audit scripts, and an incident watch list for all three moments — npm, PyPI, crates.io, Go, CI/CD, and AI agent workspaces.

AI writes. SPARDA proves. A deterministic, offline gate that catches when an AI edit removes a guard, exposes a route, or breaks an invariant — no API key, right in the agent edit loop.

Let AI agents (claude code, cursor, gemini cli, codex) send sms messages on your behalf using your actual phone via bluetooth

Self-hosted uptime monitoring on Cloudflare Workers with public status pages

Drop-in AI workflows for your app

Open-source tools for bounded AI-agent handoffs: explicit file packets, evidence receipts, controller-side verification, and a native Hermes plugin.

An autonomous, multi-agent CI/CD pipeline that daily orchestrates, generates, and validates highly technical AI skills. Powered by Groq and governed by strict rate-limiting architectures.

Action Commit Safety for AI agents: the approved tool call is the one that executes, and uncertain side effects are never blindly retried. Framework-agnostic Python runtime with durable approvals, idempotent execution, and signed audit.

One evidence-driven Agent Skill for discovering raw ideas into a Product Brief or auditing formed approaches.

Production-ready OpenCode + Oh-My-OpenCode configuration pack. 7 AI providers, 12 specialized agents, git safety and worktree plugins, plus a wisdom management system. Reusable presets, skills, and setup examples for AI coding workflows. One-command install with automatic backups.

An orchestrator that runs coding agents unattended — and never believes them. Event-sourced, independently gate-verified, resumable from any crash.

Spec-driven AI development workflow for the terminal — from idea to reviewed merge.

Agent Lifecycle Framework: Documentation as executable infrastructure for birthing and evolving autonomous AI agents

The code knowledge graph that reviews your changes. Blast radius, risk memory from git history, and 82-94% fewer prompt tokens. Zero Docker, no LLM to build, 11 languages

Toolkit for keeping Claude Code and Codex instructions, skills, and hooks synchronized.

Skills and C# command recipes for AI agents controlling the Unity Editor via MCP

Planoda — the AI-native work platform that replaces Linear, ClickUp, Monday & Trello. Issues, boards, roadmaps & automations on one schema, with AI agents that propose and humans who approve. Immutable audit log, per-workspace AI cost ledger, Postgres RLS, SSO/SAML/SCIM. Try the live demo — no sign-up.

Local-first AI code provenance: line-level attribution, PR risk reports, policy checks, and tamper-evident audit trails for AI-assisted software.

A GPU-accelerated terminal for agentic coding with native Git worktrees and diff review.

CLI tool to manage Claude Code hooks — creates, registers, and tracks prompt and script hooks without manually editing settings.json

Hands-on lab for demonstrating and reporting OWASP Top 10 for LLM vulnerabilities in AI agents.

llmwiki (Quiz_wiki) — auto-captures your coding-agent sessions (Claude Code / Codex / OpenCode) into a compounding, local-first project wiki that quizzes you back. The llmwiki quiz is a daily forgetting-curve quiz (/wiki-quiz) that keeps your memory of your own decisions sharp. Markdown as source of truth, no MCP, no build.

The project allows you to manage isolated containers with AI agents

Agent skills for trapstreet.run — set up the tp CLI, build solutions against an eval task, and author new tasks, from plain language. Claude Code, Cursor, Codex and 70+ more.

Flight search and booking Skill for AI agents. Automatically installs the atlas-flight CLI.

Visual configuration editor for OpenCode AI coding agent. Configure settings, manage plugins including Oh My OpenAgent, import/export as JSONC. Dark blue/purple theme, self-host via Docker.

Visual workflow automation platform for .NET with AI agents, webhooks, plugins, and a Blazor node designer.

Give your AI agent a WhatsApp account — an MCP server with 87 tools for reading, sending, groups, status, scheduling & more. Works with Claude, Codex, Cursor, OpenCode & any MCP client.

Open-source no-code platform for building and running AI-powered workflows — drag, drop, automate.

Verify what your AI coding agent actually did. Deterministic local checks, no LLM in the trust path. By Swail Labs.

Optimized full-stack starter repo for usage with Coding Agents

Attention triage for herdr — ranks your agents by who needs you most; a long-blocked agent rises to the top.

Portable Agent Skills for disciplined software delivery, learning, editing, and LLM evaluation

A harness for Claude Code, for people whose job is not code

Read text out loud with a good neural voice (Kokoro), from anywhere on your Mac — so an agent can get your attention or give a spoken update.

mcp de godot totalmente profesional con todas las herramientas posibles para tareas basicas, avanzadas y dificiles. sin nada malicioso. para todos el mcp que cree.

Artemis: a local-first visual canvas for running Claude Code, Codex, Gemini, and OpenCode coding agents in parallel git worktrees, with human approval gates before anything merges — no telemetry.

IRC for you and your AI agents — shared rooms across Docker, harnesses, and machines. Single Go binary with MCP, HTTP/CLI, and a web UI.

Let local AI agents operate final remote Linux environments through a tmux pane, even behind bastion hosts, jump servers, MFA, and private networks.

Enterprise orchestration engine for Agentic AI. Instantly deploy highly-scalable, headless visual sandboxes for full computer automation.

AI-powered agentic framework for web application penetration testing.

A plugin for sidebar chatting with Google Gemini and Deepseek chatbot in Zotero reader

Leave GPT4-generated comments by specifying Telegram channels to monitor & prompts to customize behavior

LawPal is India's advanced AI-powered legal assistant, designed to make legal knowledge accessible to everyone, free of cost.

Private, local-first library for your saved Instagram content with an AI chatbot — powered by Ollama, exports to Obsidian

Companion project for my course "AI Engineering Fundamentals - Build and Deploy real LLM apps with Python"

Podcast transcript API and MCP server — any published podcast as clean Markdown with real speaker names. For Claude, Cursor and other AI agents.

Unified ingestion tool for Google's Open Knowledge Format (OKF): validate, load into a portable DuckDB catalog, and semantically search bundles — from R or Python.

Lightweight HTML-to-Markdown parser for AI agent workflows — Python CLI and API around a fast Go engine, available on PyPI.

⚡ FlowGame execution engine focused on business workflows. 专注于业务落地的工作流执行引擎，支持同步/流式试运行、Redis 流程存储、Qdrant 向量检索等场景。🐍 FastAPI · 🔗 配套 FlowGame 前端