This daily update tracks GitHub repositories experiencing the highest relative momentum, ranking projects by percentage star growth over the last seven days. By prioritizing relative gain over absolute counts, we identify emerging tools like Google’s Elephant Goldfish Model—which saw a 3,580% surge—rather than just established industry giants. This week’s data highlights a concentrated interest in specialized agentic workflows, from the 3D modeling automation of Blender Skills to the local-first orchestration provided by OpenBot and the security-focused proxying of ThinkWatch Lite.

#RepositoryStars GainedGain %Overview
1google/elephant-goldfish-model59073580%An open-source plugin for coding agents that uses specialized AI roles to brainstorm, draft, and stress-test technical specifications before implementation.
2nightly-labs/openbot3187412.8%A local-first desktop workspace for managing persistent AI teammates across multiple providers like Claude, Gemini, and Grok.
3arjun988/blender-skills1144299.5%A comprehensive library of 94 specialized skills that enables AI agents like Claude Code and Cursor to execute complex 3D modeling, texturing, and animation workflows directly within Blender via MCP.
4ThinkWatchProject/ThinkWatch-Lite277236.8%A local gateway for AI clients like Claude Code and Cursor that provides security filtering, cost tracking, and seamless switching between different LLM providers.
5C4rohan/Lampline379231.1%An open-source framework for building, testing, and running self-hosted voice agents with support for Dialogflow CX and various LLM providers.

Elephant Goldfish Model (google/elephant-goldfish-model )

The Elephant Goldfish Model (EGM) is an open-source plugin designed to enhance the software design process through multi-agent collaboration. It utilizes an ‘Elephant’ agent to maintain project context and draft specifications, while ‘Goldfish’ subagents perform zero-context reviews to identify unstated assumptions and implementation risks. Developers integrate EGM into existing workflows using tools like Claude Code or Codex to brainstorm ideas, generate design documents, and run drift reports. This structured approach ensures architectural trade-offs are resolved and specifications are rigorously vetted before any code is written.

OpenBot (nightly-labs/openbot )

OpenBot is a desktop workspace designed to host persistent AI agents with individual workspaces and conversation histories. It supports a wide range of providers, including Claude Code, Gemini, and Cursor, while offering local-first data management and agent-to-agent messaging. Users can leverage features like per-agent context monitoring, embedded browsers, and computer use capabilities to automate complex workflows. This tool is primarily intended for developers and power users seeking a centralized, privacy-conscious environment to orchestrate multiple AI agents.

Blender Skills (arjun988/blender-skills )

Blender Skills provides a massive suite of 94 specialized instruction sets designed to transform AI coding agents into autonomous technical artists. By leveraging the Model Context Protocol (MCP), the repository allows users to automate entire production pipelines, including hard-surface modeling, character rigging, and environment design, directly within the Blender interface. It is specifically built for 3D artists and game developers who want to use tools like Cursor or Claude Code to execute complex tasks—such as matching reference images or optimizing assets for Unreal Engine—without manual UI interaction. The system uses an orchestration layer to route high-level requests through specific technical workflows to ensure production-standard outputs.

ThinkWatch Lite (ThinkWatchProject/ThinkWatch-Lite )

ThinkWatch Lite acts as a local proxy gateway for AI development tools such as Claude Desktop, Cursor, and Aider. It enables users to switch between various LLM upstreams—including OpenAI, Anthropic, and Amazon Bedrock—without reconfiguring individual clients. The software provides critical security features by scrubbing sensitive data like API keys and bank card numbers before requests leave the machine, and it can intercept dangerous tool calls. Additionally, it offers granular observability by recording request costs, routing paths, and comparing model performance across different providers.

Lampline (C4rohan/Lampline )

Lampline is an open-source platform designed for developing and testing automated phone agents. It enables users to build voice agents using YAML and Markdown, facilitating version control and automated testing via the vk CLI. Key features include the ability to import existing Dialogflow CX agents, run AI-driven call evaluations, and deploy on-premise to maintain data privacy. It is particularly useful for developers and enterprises looking to integrate voice AI with Twilio, Telnyx, or SIP trunks while maintaining control over model costs and hosting.