A .recur/ folder would solve most of the gaps in one shot:
.recur/
├── config.toml # Project-specific settings
├── checkpoints.md # Default checkpoint log location
└── (future: cache, indices)
# .recur/config.toml
[docs]
dir = "docs/"
sep = "."
[src]
dir = "User Level Up/"
sep = "."
[checkpoint]
file = ".recur/checkpoints.md"
root_pattern = "**" # not hardcoded to "main.*"
[status]
# recur status reads these
current_suffix = ".current.md"
todo_suffix = ".todo.md"
complete_suffix = ".complete.md"
That fixes:
- Lane detection —
root_pattern = "**"means checkpoint findsCreateWizard3.**.current, not justmain.** - Checkpoint file location — no more guessing,
recur-git checkpoint --append-paralleljust knows where to write - Multi-dir awareness —
docs/uses.separator,src/uses.too but different root recur status— knows which suffixes to count without flags
And it follows the convention every developer already knows: .git/, .vscode/, .github/. You recur init once, it creates .recur/config.toml, and every command after that just works without flags.
Attempt to implement where the recur-git logs to and how it works. Use recur-git commands to configure recur and recur-git to specific solution or project working on.