Technical notes on current mechanics, product boundaries, and trade-offs. Each note stays narrow: one current path, one boundary, or one replay/debugging surface at a time.
Present-tense claims stay close to the repository.
Start Here
The clearest first reads for the current operating story.
Text-firstRUST
// simplified current order on the owned patch lane
run_fs_write_pre_hooks(..., &req.action).await?;
let result = apply_patch_in_dir(...);
CI, local hooks, and an owned write path do different jobs
Similar rule logic can appear in all three places, but they are not interchangeable. The difference is who owns the write, when rejection happens, and what class of mistake each path is good at catching.
Read for: What changes when the runtime owns the mutation path instead of validating after the fact or on one workstation.
Repo config and event kinds should not be rebuilt locally
Direct reads of `.tsqoba/tsqoba.toml` and raw event-string matches look different, but both create side doors next to an interface that already owns the job.
Read for: Why local side doors around loaders and typed enums make the repository harder to reason about.
A smoke replay printed a named snapshot diff
Replay becomes useful when it keeps the snapshot path, bundle context, and a readable diff instead of collapsing everything into a flat suite failure.
Read for: How the current replay harness turns one mismatch into a concrete debugging surface.
A write-time 400-line Rust file cap
The number is crude. The useful part is that oversize is rejected on the write path before a large file becomes local gravity.
Read for: Why early size pressure can change the next move even when the rule itself is simple.
