Self-learning vector memory for AI agents — single-file .rvf cognitive container with HNSW search, episodic Reflexion memory, causal graph + Cypher, 9 RL algorithms, Thompson Sampling bandit, 41 MCP tools, hybrid (BM25 + dense) retrieval, GNN attention. 1

Two-phase DeepSeek Harness preset: Minimal-aligned bootstrap, then full Standard tools (Project2 98/99)

Local-first AI agent runtime with sandboxed sessions, MCP tools, memory, credentials, audit/replay, and a built-in console. Run OpenAI, Anthropic, MiniMax, DeepSeek V4, and OpenAI-compatible models on your infrastructure.

Self improving agents through iterations

Agent-native TypeScript framework for building MCP servers. Build tools, not infrastructure. Declarative definitions with auth, multi-backend storage, OpenTelemetry, and first-class support for Bun/Node/Cloudflare Workers.

Remote-work assistant for DeepSeek Harness (DSH): connect SSH (key or password), pick a remote workspace, operate with rw_* tools, and SFTP-mirror it into a real local DSH workspace.

A caring memory companion for DSH — three-layer auto memory, proactive calendar reminders, warm AI greetings, per-turn auto-consolidation, and inheritance of memories from other AI tools.

Enable AI agents to work reliably - giving them secure access to structured data, tools to take action, and the context needed to make smart decisions.

Unified notification push plugin for DeepSeek Harness (DSH): one minimal notify() API, 8 channel adapters (telegram/dingtalk/feishu/wxpusher/pushplus/serverchan/bark/webhook), dual trigger (auto session events + agent tool).

Open-source GRC toolkit from the GRC Engineering Club. Claude Code plugins for evidence collection, SCF crosswalks, multi-framework gap reports, OSCAL workflows.

A toolkit of skills, templates, and guides for bootstrapping AI-agent-friendly projects. Started from Anthropic's Claude Code engineering patterns, extended with a per-project self-improvement loop and more.

MCP server for full Godot 4.x engine control: 157 tools for AI-driven game development (GDScript and C#/.NET). Tested with Godot 4.7.

Safari browser automation for AI agents — native macOS, zero Chrome overhead. 97 tools via AppleScript + JavaScript.

Local-first AI agent runtime with sandboxed sessions, MCP tools, memory, credentials, audit/replay, and a built-in console. Run OpenAI, Anthropic, MiniMax, DeepSeek V4, and OpenAI-compatible models on your infrastructure.

Model-driven context management (Active Context Pruning / ACP) for the DeepSeek Harness — the model decides when and what to compress. Ported from billion-context-pi (ranxianglei); acp-kernel reused verbatim. CompactionEngine backend with compress/decompress/search_context/acp_status tools.

DSH plugin: pixel-to-text image reading for text-only models. image_scan/image_ocr/image_sample tools + image-reading skill (34-image trained methodology). Pure local, optional PaddleOCR.

Search & analytics data as infrastructure — MCP server for Google Search Console, Bing Webmaster Tools, Google Adsense and GA4, designed for AI agents and automation.

Bounded, layered, approval-gated, auditable cross-session memory for DeepSeek Harness (capability seam: ctx.memory + SQLite provider + memory tool + frozen snapshot injection)

Deploy static websites from AI agents. Start with the hosted endpoint at mcp.shipstatic.com — no install, no signup, no API key — or install this package to deploy folders from your own machine. The same fifteen tools either way.

AI made building cheap. It didn't make deciding cheap. Mycelium is a Claude Code harness that makes your agent run discovery and weigh evidence before it writes code. It earns the right to start. Built for software, courses, AI tools, and services.

Cross-instance message/event handoff plugins for DSH (interconnect service + tools)

Contabo API (v1.0.0) as MCP tools for cloud provisioning, and management. Powered by HAPI MCP server

MCP server that lets AI agents (Claude Code, Claude Desktop, Cursor, Copilot, Codex) build, run, inspect, and publish browser extensions. 30 tools for scaffolding, live DOM inspection, log streaming, and store-ready builds across Chrome, Edge, Firefox, Sa

Claude Code-style declarative permission rules for DeepSeek Harness: ordered allow/deny/ask rules with tool-name, argument (glob/regex), and workspace-path matching on the tools/pre-execute waterfall, session-log audit, and HMR reload.