Invisible Dots is a self-hosted, open-source platform for deploying and managing persistent AI agents ('Dots'). Each Dot operates within its own dedicated QEMU virtual machine, providing a secure, isolated environment with a full Linux desktop, browser, and file system. The platform is designed for undetectability and long-running autonomy, serving as an open-source alternative to proprietary agent services. The system is a polyglot monorepo with a strict separation between the host control plane (TypeScript) and the guest agent (Go) and engine (Python). The host manages VM lifecycle, scheduling, a web UI, CLI, and messaging channels (Telegram/WhatsApp). The guest runs a Go daemon for low-level system access and a Python engine (forked from nanobot) that orchestrates LLM calls, tool execution, and permissions. Key features include a granular permission system with approval workflows, a cron-like task/automation system, cost capping per task, and a comprehensive benchmark suite for evaluating agent capabilities. The codebase demonstrates exceptional engineering craftsmanship with clean abstractions (Database, Channels, Computer), robust testing strategies (contract tests, fakes, e2e), and reusable utility libraries (ISO writer, SSE parser, atomic file operations). While primarily an application, the project includes a significant research-oriented benchmark suite (tests/bench) and the engine itself is a research fork (nanobot), giving it a hybrid character. The core value, however, is as a production-grade application platform.
planning-with-files is a persistent, file-based task-planning system for AI coding agents. It maintains task_plan.md, findings.md, and progress.md on disk and wires into agent lifecycle hooks (SessionStart, UserPromptSubmit, PreToolUse, PostToolUse, PreCompact, Stop) across Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, Pi, Hermes, and a dozen other assistants, re-injecting plan context into the model's context window on every turn. Plans survive context loss, /clear, compaction, and crashes; a session-catchup mechanism recovers prior transcripts. A v3 gated mode can block the agent from stopping until plan phases complete, backed by an append-only JSONL run ledger, stall detection, and SHA-256 plan attestation that makes hooks refuse to inject tampered content. Architecturally it is a distribution hub rather than a single application: a canonical skill under skills/planning-with-files is mirrored byte-identical into 15+ IDE-specific directories, with native TypeScript plugins (OpenCode, Pi, DSH), a Python plugin (Hermes), and shell/PowerShell hook dispatchers. Plan resolution is a fail-closed chain (PLAN_ID env → .active_plan pointer → newest plan dir → legacy root) guarded by slug validation, containment checks, symlink/junction rejection, and strict PWF_PLAN_ROOT pinning. The engineering is unusually rigorous for an agent-skill repo: 400+ pytest tests, byte-parity tests between shell/PowerShell/Python/TypeScript twins, a PATH-stub BSD userland simulation to catch GNU-only flags that silently fail on macOS, and heavy Windows PowerShell 5.1 / Git Bash portability hardening. Non-obvious patterns include KV-cache-stable injection output (no timestamps or free text in summaries, wall-clock normalization to T00:00:00Z) and an mtime-keyed attestation cache for performance. The repository is a developer tool, not a data/research project; although it ships benchmark and eval documentation (assertion pass rates, recovery-turn comparisons), those are verification evidence for the skill, not analysis artifacts. It is actively maintained at v3.24.0 with recent commits, 4 open issues, and 100+ releases; technical drift is low, with the main maintainability risk being deliberate duplication across 20+ canonical mirrors, mitigated by sync-ide-folders.py and parity tests but still a possible source of silent drift.
Nanobot is a self-hosted, open-source personal AI assistant framework that operates as a full-stack application. The Python backend provides an agent loop with tool execution, session management, multi-provider LLM orchestration (OpenAI, Anthropic, etc.), and a plugin system for 18+ messaging channels (Telegram, Discord, Slack, WeChat, Feishu, etc.). It exposes a WebSocket-based gateway, an OpenAI-compatible API, and a CLI. Two frontends exist: a rich WebUI built with React/TypeScript/Vite/Tailwind and a terminal TUI built with Bun/OpenTUI. The architecture is event-driven via a message bus, with decoupled channel adapters, provider abstraction, and a config system supporting hot-reload. It includes security features like workspace sandboxing, credential redaction, and SSRF protection. The codebase is extensively tested (hundreds of unit/integration tests) and demonstrates strong engineering practices with clean abstractions, though its large dependency surface and multiple frontends introduce some maintenance surface area.
This repository is a comprehensive, practitioner-focused educational hub for generative AI. It is not a deployable software application or a traditional data pipeline, but a curated knowledge base that bridges theory and practice. The content is organized into free courses (Applied LLMs Mastery, Agentic AI, AI Evals), interview preparation guides for multiple AI roles (AI Engineer, AI PM, FDE, AI Strategist), system design case studies with runnable LangGraph code, and curated research updates (monthly paper lists, State of AI reports). The architecture is a hybrid of educational content and reusable code patterns, with the core value lying in its structured learning paths and conceptual frameworks. The repository's dominant character is educational content curation and research tracking. The code examples are explicitly demonstration-grade, designed to teach production-adjacent patterns like provider-agnostic LLM layers, hash-chained audit trails, and deterministic guardrails. These patterns are genuinely reusable and demonstrate strong engineering craftsmanship. The information architecture of the repo itself is excellent, making it easy for learners to navigate from foundational concepts to advanced system design. The repository is actively maintained with a massive community following. Its primary risk is the inherent rapid obsolescence of the AI field, which the maintainers mitigate by explicitly archiving outdated content and providing clear versioning. The code examples are not production-ready out of the box, but the patterns they teach are directly applicable to real-world systems. The research paper lists and interview questions require ongoing curation to stay current.
Vigil is an open-source AI SOC (Security Operations Center) platform that orchestrates autonomous threat hunting, incident response, and case management through a multi-agent architecture. The system ingests security findings from SIEMs (Splunk, Elastic, Azure Sentinel), EDRs (CrowdStrike, Microsoft Defender), and other sources via Kafka or webhooks, then uses LLM-driven agents (hunt, root_cause, adjudicate, investigate, compose, chat) to analyze, enrich, and respond. A TypeScript agent layer with event-sourced ledgers and budget enforcement runs investigations, while a Python daemon handles polling, deduplication, and orchestration. The platform exposes its own tools via MCP (Model Context Protocol) and integrates with 30+ external security tools through a descriptor-based plugin system. The frontend is a React SPA with a setup wizard, dashboards, and case management UI. Deployment is via Docker Compose or Helm, with PostgreSQL, Redis, and optional Kafka as backing services.
This repository is a production-grade, AI-powered stock analysis platform for CN, HK, US, JP, KR, and TW markets. In production it runs a scheduled analysis pipeline that fetches market data from multiple providers (akshare, yfinance, tushare, efinance, futu, etc.) with priority-based fallback, feeds the data to LLM backends via LiteLLM, generates structured decision reports, and pushes them through 15+ notification channels (WeChat, Feishu, DingTalk, Telegram, email, Discord, Slack, etc.). It also exposes a FastAPI REST API, a React/TypeScript web dashboard, an Electron desktop client, and bot integrations for Feishu/DingTalk/Discord. The system includes agent-based chat with tool use, screening, backtesting, alerts, portfolio management, decision-signal tracking, and market review capabilities.
Morphic is a self-hostable AI answer engine and chat application. In production it streams LLM research results into a rich chat UI: it runs an agentic researcher loop that calls search tools (SearXNG, Tavily, Brave, Exa, Firecrawl), fetches and extracts page content, allocates numbered citations, and emits a JSONL-based 'spec' stream that a generative UI renderer turns into structured cards, grids, headings, buttons, and inline images. Alongside the core chat it ships usage-budget metering, guest/anonymous chat, file uploads to S3-compatible storage, a notes/library feature, feedback collection, model discovery and selection, chat sharing, admin invitations, and an analytics relay that forwards scrubbed PostHog events without exposing authentication tokens.
QueryChat is a dual R/Python open-source library that turns tabular data into a natural-language chat interface: users ask questions in plain English, an LLM generates SQL via tool calls, and the SQL executes against an in-memory DuckDB engine, a SQLAlchemy-managed database, an Ibis backend, a Polars lazy frame, or a pins board. A defining safety property is that raw data is never sent to the LLM — only schema introspection (column names, types, ranges, categorical values) and query results flow through the model. The package ships with five LLM-visible tools (get_schema, query, update_dashboard, reset_dashboard, visualize) and a hardened DuckDB mode that blocks external filesystem access and extensions, making it a well-considered solution to the inherently risky problem of letting models write SQL against user data.
MemoryBear is a production-oriented, multi-tenant AI agent platform whose core differentiator is a cognitive-inspired persistent memory system for LLM applications. The code — not the README — shows a full SaaS stack: a FastAPI backend (~60 domain controllers), a React/TypeScript admin console, an identity/auth service, an API gateway, a dedicated RAG knowledge service, and a Rust seccomp sandbox for code execution. The memory subsystem is the heart: pipelines extract statements, triplets, entities, emotions, preferences, and implicit habits from conversations into a Neo4j knowledge graph (projected to Elasticsearch through a transactional outbox), with ACT-R-inspired forgetting/reinforcement scheduling, LLM reflection/dedup, BERT scene-boundary detection, and hybrid retrieval at query time. This is an application, not a data-research repo, although the memory-science algorithms (forgetting curves, emotion classification, value ranking) are shipped as product features rather than research artifacts.
This repository, ai-hedge-fund, is a production-grade simulation engine for running AI-managed hedge funds in educational and research contexts. It provides a dual-mode system supporting both historical backtesting and live paper trading, all adhering to point-in-time data integrity to prevent look-ahead bias. The architecture is modular and protocol-based, with swappable components for data acquisition (Financial Datasets API), LLM providers (Anthropic, OpenAI, DeepSeek, Google, xAI, Kimi, TypeSafe/Jev), broker execution (simulated/paper), signal generation (quant models and LLM agents emulating investors like Buffett, Munger, Graham), and risk management. The engine includes a hash-chained ledger for tamper-evident record keeping, prompt caching to eliminate redundant LLM costs during backtests, and a textual TUI for interactive management. It is explicitly not intended for real trading and is released under MIT license. The codebase is well-structured with clean abstractions, though it carries some technical debt from being a ground-up v2 rebuild that is not yet fully integrated with its predecessor, and contains hardcoded constants and magic numbers that limit configurability.
Mobilerun is a comprehensive CLI and Python SDK framework for controlling Android and iOS devices through natural language commands powered by LLM agents. It orchestrates multi-agent workflows (Manager, Executor, FastAgent) built on llama-index Workflows, with device interaction via ADB (Android) and a separate iOS binary (WebDriverAgent). The system supports multiple LLM providers (OpenAI, Anthropic, Gemini, xAI, DeepSeek, Ollama, etc.), including custom OAuth flows for subscription-based access (Anthropic Claude, Google Code Assist, xAI Grok). It features a configurable agent architecture with Jinja2 prompt templates, macro recording/replay, MCP-based custom tool integration, telemetry (PostHog), tracing (Phoenix, Langfuse), and a rich set of developer tools (device selection, configuration wizard, doctor, credential management). The codebase is well-tested with over 50 test files covering XML parsing, coordinate coordinate contracts, OAuth flows, and agent workflows. While there is some code duplication in OAuth implementations and tight coupling to llama-index internals, the overall architecture is modular and extensible. The project is in active development (version 0.7.0, Beta) with commercial ties to Mobilerun Cloud.
AssetOpsBench is a unified open framework for building, orchestrating, and evaluating domain-specific AI agents in industrial asset operations and maintenance (Industry 4.0). It provides a suite of MCP (Model Context Protocol) servers that expose domain tools for IoT telemetry, work order lifecycle, failure mode analysis, time-series forecasting/anomaly detection, vibration diagnostics, and utility functions. The framework includes multiple agent runners (plan-execute, Claude, OpenAI, Deep, Stirrup, OpenCode, direct LLM) that consume these tools, and a comprehensive evaluation pipeline with LLM-as-judge and deterministic structured-answer scorers. The system is designed to run reproducible benchmark scenarios with ground truth, producing leaderboard reports for agent comparison. It is accepted at KDD 2026 and powers multiple competitions. Architecturally, the system is a collection of Python packages with 14 CLI entry points defined in pyproject.toml. Each MCP server is a standalone FastMCP application using stdio transport, backed by CouchDB for persistence. Agent runners share a common trajectory model and observability layer (OpenTelemetry with file-based export). The evaluation pipeline is pluggable with a scorer registry. The codebase also includes a significant data/research component: a time-series feature extraction library (130+ extractors), a model catalog with foundation models (TinyTimeMixer, sktime wrappers), synthetic data generation for vibration signals, and showcase notebooks demonstrating end-to-end model lifecycle. The hybrid nature makes it both a software benchmark framework and a research platform for industrial AI.
FigMirror is a reference-conditioned scientific figure style transfer system that iteratively generates matplotlib scripts matching the visual style of a reference figure. It operates as a multi-agent loop (Drawer/Reviewer) orchestrated by a runner backend (Codex, Claude, or Mock), with a local web UI for upload, iteration browsing, and refinement. The system is packaged as skills for both Claude Code and Codex, plus a standalone web server. While primarily an application, it has a strong research component: the prompts, aesthetic library, and evaluation benchmarks (PlotTwin-Bench) are the result of extensive empirical iteration. The codebase demonstrates high craftsmanship in its runner architecture, atomic file operations, SSE streaming, and deterministic review gates. However, it suffers from dual skill bundles (Claude/Codex) that must stay synchronized, retired artifacts, and some incomplete features (ClaudeRunner, SSE integration).
Jellyfish is an open-source, full-stack AI short drama studio application. It provides an end-to-end production pipeline for generating vertical short dramas from scripts: script parsing (division, element extraction, consistency checking), storyboard generation (shot breakdown, asset linking, camera language), image/video generation via multiple AI providers (OpenAI, Volcengine), and async task orchestration with Celery. The backend is built with FastAPI, SQLAlchemy 2.0 async, and Redis; the frontend uses React 18, TypeScript, Ant Design, and Vite. The application is production-oriented with Docker Compose deployment, GitHub Actions CI, and automated OpenAPI client generation. It emphasizes visual consistency through global seed, style inheritance, and asset reuse. While primarily an application, it includes a significant amount of AI/LLM agent code for script analysis, giving it a hybrid character with research-oriented structured extraction pipelines.
This repository is a collection of Claude/Anthropic skills and platform documentation rather than a standalone application. It contains skill definitions (SKILL.md files), MCP server evaluation guidance, managed-agents documentation, and supporting tooling such as schema definitions and build scripts. The dominant character is developer-facing content and reusable patterns for building LLM agents, with some executable scripts for validation and report generation. It provides structured output schemas for evaluation pipelines (grading, benchmark, comparison, analysis), a validator hierarchy for XML/XSD-based skill files, and a security-conscious file I/O example in build-report-lite.mjs. It also documents managed agents, platform availability, and MCP best practices. The repository is not a data research project; there are no notebooks, data pipelines, or model training scripts. Maintenance is active, but technical drift is medium due to version-sensitive SDK references, manually maintained platform availability tables, and multiple language-specific examples that must be kept in sync. The code and schemas are generally well-structured and reusable, though some paths are hardcoded and some documentation is tightly coupled to generator scripts.
OASIS is a scalable, open-source framework for simulating social media platforms (Twitter, Reddit) using LLM-powered agents. The system models up to one million agents with realistic actions (post, like, follow, comment, report, etc.) and supports configurable recommendation systems, agent graphs, and time scaling. It is designed for research into information spread, misinformation, group polarization, and herd behavior, as evidenced by extensive experiment scripts and real-world datasets included in the repository. The architecture follows a PettingZoo-style environment loop with a Platform, SocialAgent, AgentGraph, and RecSys modules, all built on top of the CAMEL framework. While the core is a reusable simulation library, the repo also contains research data, visualization scripts, and one-off experiment configurations, making it a hybrid of software application and research project.
Pentest-Swarm-AI is an open-source autonomous penetration testing tool built on a stigmergic swarm architecture. It coordinates multiple LLM-driven specialist agents (recon, classifier, exploit, report) via a shared blackboard with pheromone decay, enabling emergent attack paths without a central planner. The system integrates 30+ external security tools (nmap, nuclei, sqlmap, etc.), multiple LLM providers (Claude, OpenAI, Ollama, Gemini, LM Studio), and bug bounty platforms (HackerOne, Bugcrowd, Intigriti) for automated scope import and report submission. The architecture is a multi-agent system with a scheduler that dispatches agents based on blackboard predicates and pheromone levels. Key entry points include a CLI (Cobra commands: scan, init, playbook, submit, etc.), an API server (Fiber), a WebSocket hub for live streaming, a Next.js web dashboard, and an MCP server for IDE integration. Data flows: agents write findings to the blackboard with pheromone weights; the scheduler reads the blackboard and triggers agents whose predicates match; agents read findings, process them, and write new findings, creating a stigmergic feedback loop. External dependencies include PostgreSQL with pgvector for durable blackboard storage, LLM API endpoints, and the installed security tool binaries. The codebase demonstrates high craftsmanship with clean abstractions (tool adapter pattern, LLM provider interface, blackboard interface), extensive testing (unit, integration, e2e, benchmarks, LLM eval), and safety features (scope enforcement, provenance signing, rate limiting, false-positive caching). It is actively maintained with recent commits, CI/CD pipelines, and community engagement.
ClawBench is a rigorous execution-first benchmarking framework for evaluating AI agents (particularly OpenClaw) on multi-step tasks. The system orchestrates agent runs through a gateway or containerized native evaluation, scores performance across four axes (completion, trajectory, behavior, advisory judge), and generates detailed diagnostic reports. Its architecture includes a job queue with HuggingFace Dataset persistence, a Gradio web app for interactive leaderboards, and a comprehensive CLI for running evaluations and analyzing results. The framework supports multiple agent runtimes via a pluggable adapter pattern (OpenClaw, Hermes, Claude Code, Codex) and provides a canonical task schema for agent-agnostic task definitions. Beyond basic scoring, ClawBench applies dynamical-systems and statistical methods to agent behavior: trajectory regime classification (trapped, limit-cycle, diffusive), Kaplan-Meier survival analysis, constraint index computation, and variance decomposition. It includes a configuration-space benchmarking system (v0.5) that fingerprints plugin profiles for cold-start prediction and fANOVA factor analysis. The codebase is well-tested with over 60 test files, uses Docker for reproducibility, and includes CI/CD pipelines for automated evaluation campaigns. However, the system is complex and has some hardcoded paths and environment variables tied to specific deployment contexts (Crabbox, OpenClaw gateway), and its heavy dependency on external services introduces maintenance risk. Overall, it is a production-grade benchmark with sophisticated analysis capabilities.
This repository is the NVIDIA BioNeMo Agent Toolkit — a collection of 31 skills that give coding/scientific agents (Claude Code, Codex, any SKILL.md-compatible harness) domain knowledge for life-science workflows: protein structure prediction, molecular docking, generative chemistry, and related tasks. Skills execute in the user's own environment (workstation, notebook, cluster) and wrap NVIDIA NIM (NVIDIA Inference Microservices) APIs — both hosted endpoints and local Docker deployments. The toolkit is not an installable Python package; it is a skill catalog with supporting tooling, dual-licensed (Apache-2.0 OR CC-BY-4.0).
This is an 'awesome list' curation repository — the definitive index of the LangGraph/LangChain ecosystem — rather than a conventional software application. The core product is a ~30k-line README.md that maps the ecosystem into core frameworks (LangChain, LangGraph, Deep Agents, LangSmith, LangSmith Fleet, and integrations), official LangChain projects, and ~100 community projects organized across 20+ use-case categories (RAG, finance, web automation, coding agents, research, cybersecurity, workplace productivity, media, robotics, and more), plus learning resources, templates, case studies, and communities. Its practical function is as a discovery and evaluation map for engineers building stateful, tool-using AI agents: everything is categorized, linked, badged, and maintained under defined contribution quality gates. The only real code is scripts/repo-checks.mjs, a maintenance automation script that parses every GitHub URL from the README, queries the GitHub API for each repo's metadata, classifies repos as archived or stale (15-month staleness cutoff for community projects), writes a markdown check report, and fails CI when problems are found. This is orchestrated by three GitHub Actions workflows: awesome-lint/markdownlint on PRs, a monthly scheduled repo-health cron, and a stale-issue/PR bot. The package.json exposes npm scripts for these checks with Node >=20 as the engine. The repository is actively maintained (latest commit 2026-07-10), uses current GitHub Actions major versions (v6 checkout/setup-node, stale@v9), and follows a documented contribution process with a PR template enforcing category placement, alphabetical order, and non-archived/active-project requirements. For AI agents and engineers, the value is twofold: (1) it is a currently-maintained, comprehensively categorized starting point for discovering LangGraph/LangChain projects and patterns by domain — useful before writing any code; (2) the repo-checks script is a genuinely extractable pattern for anyone maintaining any markdown document containing GitHub links, since it is a self-contained, env-configurable health checker that requires minimal adaptation. It is as close to a 'productized' awesome list as exists in this space, with CI, lint enforcement, and automated freshness audits. Limitations: no unit tests for the maintenance script (its errors surface only monthly via cron), descriptions for some official projects are generic and low-signal, and there is no actual application or library code to reuse beyond the single maintenance script.