Repo: https://github.com/soad666p/Rubber_Ducky
MCP server that can be used from any MCP-compatible client (for example Gemini CLI, Claude Desktop, or other LLM tools). It exposes workflow tools (Jira, Confluence, GitHub, Splunk, SonarQube, etc.) and guides through proposals, planning, implementation, testing, log analysis, and PR creation.
You can install this as a Gemini CLI extension directly from GitHub:
gemini extensions install https://github.com/soad666p/Rubber_DuckyNote: If running locally, you must manually run npm install in the extension directory.
npm install
npm run buildTo run the MCP server directly (e.g. for debugging):
npm run devnpm testThe AI Rubber Duck now acts as an active mentor, not just a code generator:
- Socratic Questioning: Asks "why" before "what" to trigger analytical thinking
- Source Citations: Always cites specific standards and ADRs
- Tool Discovery: Proactively suggests relevant tools per persona
- Psychological Safety: Non-judgmental, curious, teaching-oriented tone
Duck_Context_Switch— Context switching with Jira/Confluence briefingDuck_Security_Audit— Security-focused code review with CVE checksDetect_Network_Hopping— Network hopping violation detection in testsDuck_Onboard_Codebase— Architecture overview and onboarding guide
Make_TODO_INTO_JIRA_TICKETS— Convert TODOs to Jira ticketsExplain_the_codebase— Detailed code explanationsfull_sdlc_from_jira_to_github— End-to-end SDLC workflowCheck_for_test_standards— Test pyramid compliance auditReport_the_latest_test_failures— CI failure analysis- And more...
This project is submitted to AtlanTec AI Challenge 2026 under the Innovation category.
Issues Implemented:
- #12 - AI Active Mentorship voice & tool discovery (€1000)
- #13 - Unit & integration tests for new Duck tools (€1000)
See repository license file.