open-source

CodeGraph is a local code knowledge graph for AI coding agents like Claude Code, Cursor, Codex, OpenCode, and Hermes Agent. It aims to cut token use, tool calls, and runtime by letting agents query pre-indexed code structure instead of scanning files repeatedly.

Open-source security scanner for AI agents, MCP servers, and agent skills. It auto-discovers installed agent components and checks them for prompt injection, tool poisoning, secrets, malware payloads, and related risks.

An open-source SAS 9.4-compatible interpreter written by agents, with a browser playground, tests, and real clinical-programming pipeline support.

Open-source security agent framework for cloud, code, and runtime. It ships 45 built-in agents for AWS, GCP, Azure, Kubernetes, GitHub, and GitLab, and lets users create custom agents in a single markdown file.

An open framework for scoring how well an MCP tool definition communicates to an AI agent, with a spec, reference implementation, CLI, and hosted API.

Open-source LLM gateway that adds cost visibility, request tracing, repeated-call detection, and basic prompt-injection heuristics for OpenAI and Anthropic traffic.

A pure C runtime for Gemma 4 E2B CPU inference, with export, benchmarking, and numerical validation tools included.

An OWASP incubator project that protects AI agent memory from prompt injection, secret leakage, and tampering. It includes a Python library, policy-based controls, benchmarks, and integrations for agent frameworks like LangChain and AutoGen.

A code editor for the agents era that orchestrates multiple CLI coding agents in parallel across isolated git worktrees, with terminal, review, and editor handoff workflows.

A local, offline AI assistant that runs on Mac or Windows and answers using a built-in emergency/reference library with source citations.

An open-source MCP and CLI tool that lets AI agents read and control the iOS Simulator, build and test Xcode apps, and handle common iOS interactions with a compact screen representation.

Nines is an open-source SLO platform for defining critical user journeys as code, measuring reliability with Prometheus or Mimir, and managing workflows through a web UI, CLI, Go SDK, and MCP tools.

An open-source coding agent that aims to use fewer tokens and less memory while supporting multiple models, browser automation, subagents, and MCP plugins.

An open-source Google Research framework for recursive agent harnessing, designed to improve agent behavior through task-level and harness-level recursion without changing model weights.

An open-source code-review training repo with 24 AI-written bug katas in Python and JavaScript. It includes runnable tests, fix-and-verify workflows, and GitHub Actions grading for forks.

Open source inference engine for agents that self-tunes on the user’s hardware to run open models faster, with desktop app support and one-click connections to popular agents.

A local-first agent harness for running 2–9B models on laptops with Ollama. The project focuses on finishing real tasks, includes an offline mode, benchmarks, and a large public test suite.

Open-source native macOS reading workspace for PDFs, EPUBs, and TXT files with highlights, notes, inline rewriting, book chat, browser panes, and local-first storage.

Open-source identity-based remote access platform built on WireGuard, combining reverse proxy, VPN, SSO, and access control. The page also shows an AI gateway use case for putting cloud and self-hosted model access behind a single authenticated URL.

Open-source platform for building internal tools, dashboards, business apps, and workflows, with AI app generation and MCP support for coding agents.

A PostgreSQL helper that previews INSERT, UPDATE, and DELETE effects as row-level proposals before applying them, aimed at AI agents and other write workflows.

Podiom is a local-first orchestration layer for Claude Code and OpenAI Codex that keeps durable sessions, agent profiles, scheduling, and goal tracking in one workspace.

Knowl is a persistent memory server for AI coding agents. It stores facts with provenance, supports supersession when information changes, and integrates with clients like Claude Code, Cursor, and Codex through MCP.

Molt is an open protocol for letting an AI agent make bounded purchases through disposable, single-use payment credentials. The repo includes a clear quickstart, MCP integration, a threat model, and a working test-mode demo.