This guide configures a speed-first commit workflow with deferred deep review:
- Fast lane now: commit immediately with low-latency model fallbacks (GLM/Cerebras via
zerg/ai). - Deep lane later: batch Codex reviews across queued commits.
[commit]
queue = false
queue_on_issues = false
message_fallbacks = [
"rise:zai:glm-5",
"rise:cerebras:gpt-oss-120b",
"remote",
"openai"
]
review_fallbacks = [
"glm5",
"rise:cerebras:gpt-oss-120b",
"codex-high"
]
[options]
# Optional: mirror commit + queued-review results to myflow.
myflow_mirror = true
# myflow_url = "https://myflow.sh"
# myflow_token = "..."
# Optional: route Codex review through a wrapper transport binary.
# Must implement `app-server` JSON-RPC compatibility.
# codex_bin = "~/code/flow/scripts/codex-jazz-wrapper"# Fast default commit path (quick lane)
f commit
# Later, run deep Codex review across backlog
f reviews-todo codex --all
# Inspect pending/updated deep-review todos
f reviews-todo list
# Approve once issues are addressed
f reviews-todo approve-all- Each reviewed commit writes a report under
~/.flow/commits/. - Use that report directly:
f fix ~/.flow/commits/<report>.mdThis replaces manual “copy f commit output into Codex and ask to address all issues”.
- Plain
f commitalready uses the fast lane (--quick) by default. Setquick-default = falseonly if you want blocking review as the default. - Async queued Codex reviews now emit
commit_queue_reviewmirror sync events to myflow/gitedit when the reviewed commit is currentHEAD(the defaultf commit --quickflow). f reviews-todo codex --allis a workflow alias over commit queue deep review.