Forge produces inferior results compared to other agentic solutions out there

[sirmacik]

Hi,

I value Forge’s UI, philosophy, and features. Compared with tools like SST/Opencode, Claude Code, or Kilo Code, I notice Forge often needs more time and tokens to reach a solution (e.g., starting in Muse, then switching to Forge mode).

Could you share why this happens? Is it mainly prompt-related or are other factors involved? What steps might improve performance?

I’m keen to exchange ideas and help make Forge even better.

(@tusharmath, I’ll revisit the GitHub integration soon.)

[tusharmath]

Hi @sirmacik! Thanks for the candid feedback 🙌
When you say performance, do you mean the final output is not as good as other tools? Or are the results fine, but it's slow to achieve those results?

Would be great to continue discussion here: https://discord.gg/kRZBPpkgwq