Back to References
AI Reference
Interesting AI Repositories
A compact shortlist of useful open-source AI repositories, grouped with one top-level category and practical subcategories.
Last updated: 2026-03-24
Catalog
Repository Catalog
All entries are AI repositories. Subcategories capture each repository's primary focus.
| Repository | Subcategory | Description |
|---|---|---|
| virattt/ai-hedge-fund | Autonomous Agents (Finance Research) | A simulation platform for finance-focused AI agents that pulls and normalizes market data, runs analyst-style evidence checks, and outputs structured investment hypotheses with rationale. |
| msitarzewski/agency-agents | Agent Frameworks (Agency Workflows) | A TypeScript scaffold for role-based agent teams with explicit handoff boundaries, role-specific prompts, tool adapters, and deterministic routing between coordinator, worker, and reviewer agents. |
| promptfoo/promptfoo | LLM Testing & Evaluation | A full LLM evaluation platform for config-driven test datasets, multi-provider prompt runs, metric scoring, and automated quality gates that fit directly into CI workflows. |
| anomalyco/opencode | AI Coding Agents & Developer Tooling | A coding-agent environment that binds model decisions to repo-aware execution (file edits, shell commands, review feedback), designed for iterative development and human-in-the-loop loops. |
| dmno-dev/varlock | Environment Security & Config Management | A typed configuration and secrets system that validates env schemas, enforces policy constraints, and helps prevent drift and misconfiguration during deployments. |
| thedotmack/claude-mem | AI Coding Agents & Developer Tooling | A persistent cross-session memory system for Claude Code that stores observations, plans, and context in a local database so agent workflows retain continuity across conversations. |
| bmad-code-org/BMAD-METHOD | Open-Source AI Agent Frameworks | A methodology + reference implementation for role-driven AI delivery, defining context handoff, team cadence, and governance checkpoints for agent-assisted engineering teams. |
| firecrawl/firecrawl | Data Extraction & Web Scraping | A web scraping and crawling platform that converts websites into clean, LLM-ready markdown or structured data, with APIs for single-page extraction, site-wide crawling, and search. |
| gsd-build/get-shit-done | AI Coding Agents & Developer Tooling | A structured workflow orchestrator for Claude Code that manages project planning, phase execution, verification, and context handoff across multi-session agentic development cycles. |
| pbakaus/impeccable | Open-Source AI Agent Frameworks | A compact agentic workflow toolkit that applies constrained prompting, staged task decomposition, and output schemas to keep multi-step behavior inspectable and deterministic. |
| volcengine/OpenViking | Open-Source AI Agent Frameworks | A production-oriented context system for agents that unifies memory, retrieval, and tool execution so long-running tasks maintain continuity across turns and components. |
| microsoft/markitdown | Data Extraction & Web Scraping | A Python utility from Microsoft that converts documents (PDF, Word, Excel, PowerPoint, HTML, images, and more) into clean Markdown, making unstructured files consumable by LLMs and search pipelines. |
| karpathy/nanochat | Open-Source AI Agent Frameworks | An extremely minimal experimental harness intended for LLM training and chat behavior experiments, with a compact, inspectable flow from data to generation loop. |