Forge‑Load: A 50‑Line Bash Bridge for Multi‑Model AI Workflows
Install forge‑load via curl‑pipe and use the --engine flag to route code to the desired AI CLI with a single command.
Run curl -sSL https://raw.githubusercontent.com/Kingmaker16/cli-forge/main/install.sh | bash and try forge-load react "Refactor the auth flow".
Summary
The author built forge‑load, a 50‑line bash script that aggregates a codebase and pipes it to the chosen AI CLI. It supports a --engine flag for claude, codex, and gemini, defaulting to gemini when no flag is provided. A curl‑pipe install script allows quick setup, and the README URLs were fixed to point to raw GitHub paths. Safety claims were added to clarify that the tool does not phone home, and the script underwent adversarial review by Codex to catch edge‑case bugs. The project is open‑source on GitHub (Kingmaker16/cli‑forge) and demonstrates how to split responsibilities among AI models. The result is a single command that routes code to the appropriate AI CLI with minimal friction.
Key changes
- 50‑line bash script that routes code to the chosen AI CLI
- Supports a --engine flag for claude, codex, and gemini
- Default engine is gemini when no flag is provided
- Includes a curl‑pipe install script for quick setup
- README URLs fixed to point to raw GitHub paths
- Adds safety claims to clarify that the tool does not phone home
- Undergoes adversarial review by Codex to catch edge‑case bugs
- Open‑source repository on GitHub (Kingmaker16/cli‑forge)