| name | devenv-lock |
| description | Use when a module edit or an input bump doesn't take effect, or to decide which cache layer to reset. The lock vs eval-cache decision tree for devenv repos. |
| auto_trigger | {"keywords":["edit didn't take","nothing changed","stale lock","re-lock","devenv update","rm -f devenv.lock","eval cache","refresh-eval-cache","lock file","wrong revision"]} |
The lock / eval-cache loop
Two cache layers decide whether your edit takes effect, and they reset
differently. Diagnose by asking what did I edit?
| What you edited | Layer | Reset |
|---|
this repo's devenv.nix / env.* | eval cache (.devenv/) | devenv shell --refresh-eval-cache -- <cmd>; big hammer: rm -rf .devenv |
a module this repo imports by path (e.g. repoman) | lock (devenv.lock) | rm -f devenv.lock; surgically: devenv update repoman |
| a remote input upstream | lock | devenv update <input> / devenv update |
| don't know | both | rm -f devenv.lock && rm -rf .devenv — slower, but the reliable recovery |
The genome renders the toolchain imports from the answers file, so a generated
devenv.yaml carries a path like:
repoman:
url: path:{{ repoman_dev_root }}/repoman/modules
That literal {{ }} is why .agents/skills/** (and the genome's
.agents/devenv/**) sit in the template's _copy_without_render — Copier must
ship these files byte-for-byte, never render them.
Background: lock-and-cache.md and the-lock-cache-loop.md in the devenv docs
export. The symptom→cause tree is devenv-troubleshoot; "I edited a module and
nothing changed" is devenv-module-edits.
For when in the lifecycle to rebuild vs. verify vs. commit, see the repoman skill.
