The whole loop is free.
Pay for more of it at once.
You already pay Anthropic or OpenAI for the models. Plyloft charges only for parallel work and teams, never per token.
- Builder and Challenger, Claude or Codex
- Findings on the diff and review rounds
- Pull requests, checks and shipping
- Any number of workspaces
- One card building at a time, across all of them
- Write and plan other cards meanwhile
- Everything in Free
- Cards in parallel, no limit
- Build with both on one card
- Merge when checks pass
- Ship batch
- Skill improvement proposals
- Founding price kept while you stay
- Everything in Pro, for each member
- Shared board on GitHub Issues
- Shared skills and review rules
- Seats the owner assigns by GitHub account
- Overlap warnings across teammates
Which plan fits you
An evening on your side project. You write one card, “Sessions should last the days set in Settings”, and press Start build. Claude builds it in its own worktree; Codex reads the build and finds that maxAge is in seconds. Claude fixes it, Codex finds nothing blocking, you press Ship it.
Monday morning. A bug fix, a new export and a refactor go on the board. All three start at once, each in its own copy of the repo. On the export you are unsure of the approach, so you use Build with both: Claude and Codex each build it, the other one judges, and you pick the better result.
The owner opens Settings › Team, Join or create a team, and creates one, paying for the seats: a private GitHub repository holds the member list and the shared skills. The owner gives each seat to a teammate’s GitHub account and gives the team its workspaces; each teammate gets “Invitation sent to alice. They accept it on GitHub, then join from Plyloft.” The board lives on GitHub Issues, so a card someone took reads Claimed by alice on everyone’s Board, and every Claude and Codex session loads the team’s skills.
Everything a team can do
Pro for every member, plus what you share. It all runs on GitHub and your own machines.