Forge vs The Alternatives
April 6, 2026 · View on GitHub
Honest positioning. All three tools solve overlapping problems; the right choice depends on what you value.
| Dimension | Forge | Ralph Loop | GSD-2 |
|---|---|---|---|
| Core metaphor | Native Claude Code plugin with streaming DAG + state machine | Re-feed same prompt in a while loop | Standalone TypeScript agent harness on Pi SDK |
| State model | Task DAG, lock file, per-task checkpoints, token ledger | One integer (iteration) + active flag | Full state machine in external TypeScript |
| Task decomposition | Milestone > Spec > R-number > Task DAG, adaptive depth | None. Claude figures it out from files | Milestone > Slice > Task hierarchy |
| Context isolation | Handoff + resume in new Claude Code session at 60% | Same session, context accumulates | Fresh 200k window per task via Pi SDK |
| Stop condition | DAG complete + verifier pass, or budget ceiling | --max-iterations OR exact <promise> match. Default infinite. | Budget ceiling + verification + Escape key |
| Cost controls | Per-task + session token budgets, hard ceilings | None built-in | Per-unit token ledger with budget ceilings |
| Git isolation | Per-task worktrees with squash-merge | None | Worktree isolation per slice |
| Crash recovery | Lock file + forensic resume from checkpoints + git log | None | Lock files + session forensics |
| Verification | Goal-backward verifier (existence > substantive > wired > runtime) | Whatever you put in the prompt | Auto-fix retries on test/lint failures |
| Setup | claude plugin install | Built into Claude Code | npm install -g gsd-pi |
| Lives in | Your existing Claude Code session | Your existing Claude Code session | Separate TUI harness |
| Author | Lucas Duys | Anthropic (technique by Geoffrey Huntley) | TÂCHES |
When Forge wins: you already love Claude Code and want autonomous execution without leaving it. Native plugin architecture, zero install friction, adaptive depth scoring, multi-repo coordination, backpropagation, readable source (markdown + bash + CJS).
When GSD-2 wins: you want a battle-tested harness with more engineering hours behind it, hard per-task budget ceilings from day one, and you're willing to switch to a separate TUI.
When Ralph Loop wins: you have a tightly-scoped greenfield task with binary verification (tests pass or fail), you don't care about cost, and you want the absolute minimum infrastructure.