A comprehensive development workflow for Gemini CLI — from TDD-driven implementation plans to execution, debugging, code review, and worktree management.
Code Forge transforms your development process into a structured, evidence-based workflow. It leverages Gemini CLI's Agent Skills and Sub-agents to automate the most repetitive parts of software engineering while maintaining high quality and security standards.
| Skill | Command | Description |
|---|---|---|
| Plan | /code-forge:plan |
Analyze requirements and generate a TDD-driven implementation plan. |
| Impl | /code-forge:impl |
Execute tasks one by one with TDD discipline and sub-agent isolation. |
| Fixbug | /code-forge:fixbug |
Debug with interactive upstream trace-back to requirements and plans. |
| Review | /code-forge:review |
14-dimension code review against best practices and design goals. |
| Status | /code-forge:status |
View feature dashboard, progress tracking, and project overview. |
| Debug | /code-forge:debug |
Systematic root cause debugging for any technical issue. |
| TDD | /code-forge:tdd |
Enforce Red-Green-Refactor cycle for ad-hoc development. |
| Verify | /code-forge:verify |
Evidence-based completion verification before claiming work is done. |
| Worktree | /code-forge:worktree |
Isolated feature development with automated setup and safety checks. |
| Finish | /code-forge:finish |
Complete a branch with PR creation and worktree cleanup. |
| Port | /code-forge:port |
Port features to new languages with automated batch planning. |
| Parallel | /code-forge:parallel |
Dispatch independent agents to solve multiple unrelated problems. |
One-command build and installation for your global Gemini profile:
npm run deployThen, in your active Gemini CLI session, run:
/skills reload
/commands reload
npm run uninstall:userCode Forge uses a .code-forge.json file in your project root to manage directory paths and execution preferences. See docs/CONFIGURATION.md for details.
- Context Efficiency: Heavy lifting (analysis, code generation) is offloaded to specialized sub-agents.
- Evidence-Based: No completion claims without fresh test evidence.
- Standardized: Enforces team-wide standards for architecture, naming, and testing.
- Modular: Self-contained skills that can be easily updated and shared.
Created by tercel. Optimized for Gemini CLI.