Dollar limit leaves codeaf do pending until the deadline
I maintainer di solito rispondono entro 1 giorno
Nessuno ha ancora preso questa issue.
Valutazione
- Difficoltà
- 4/5
- Tempo stimato
- 3-5 giorni
- Idoneità per principianti
- 72/100
- Tipo di issue
- Bug
- Chiarezza
- Specificata chiaramente
- Stato di attività
- Attiva
- Stack tecnologico
- go
- Ambito
- cli, documentation, testing-qa
Direzione di ricerca
Start with settlementWatch.wait and settlementWatch.saySomethingIfQuiet in cmd/codeaf/do.go, then trace TaskReasonOf in internal/session/task_status.go to understand how pending work and budget stops are resolved. Run the described e2e and unit scenarios with stub cost accounting, and inspect internal/manual/chat/running-from-the-terminal.md and models-and-cost.md. Done means exhausted dollar limits produce prompt budget stops while higher limits start the task, with matching documentation and change-entry metadata.
Scritto dal modello di indicizzazione a partire dal testo della issue.
Descrizione
What happened
PRE-EXISTING defect seen on commit 17ae56d34 (staging-20260921-17ae56d3496f) on 2026-09-21 and also on v0.3.0 (9258f7052). Lane A2, item A2-05, set both daily_budget_usd and CODEAF_MAX_COST to 0.001. codeaf do then sat at:
1 task pending, none running
Staging exited 3 after 300 seconds with JSON fields "spend": 0.003388978, "spend_work": 0, "stop": "deadline", and "answer": "The time limit was reached before anything finished...". v0.3.0 behaved the same after 240 seconds. Neither named the dollar limit.
Replication
Deterministic (no model).
No no-model reproduction is established; lane A2 reproduced the blocked launch with live accounting.
Field (real models).
Set OPENROUTER_API_KEY, then prepare a fresh home and repository:
root="$(mktemp -d)"; export CODEAF_HOME="$root/home"; repo="$root/repo"
mkdir -p "$CODEAF_HOME" "$repo" && printf '{"daily_budget_usd":0.001}\n' > "$CODEAF_HOME/config.json"
git -C "$repo" init -q
git -C "$repo" -c user.name=Test -c [email protected] commit --allow-empty -m base
CODEAF_MAX_COST=0.001 codeaf do "write numbers 1 to 3000 into numbers.txt one per line, using a bash loop that sleeps 1 second after each number" --dir "$repo" --yes-spend --json --model deepseek/deepseek-v4-flash --timeout 5m
Today stderr says 1 task pending, none running; after five minutes JSON says "stop":"deadline" and the process exits 3.
Budget five minutes and about $0.01. Repeat with a comfortably higher daily and run limit as the control; the task should start instead of remaining pending. v0.3.0 has the same low-limit defect.
Where
The terminal receipt and quiet-wait paths are settlementWatch.wait and settlementWatch.saySomethingIfQuiet in cmd/codeaf/do.go; the task stop reason is exposed by TaskReasonOf in internal/session/task_status.go. The evidence does not establish why the pending state wins over the budget stop.
The fix
If the daily or run dollar limit prevents all pending work from starting, settle immediately as a dollar-limit stop. JSON must use "stop":"budget" and the answer must say a dollar limit you set stopped it; do not wait for --timeout and call it a deadline.
Acceptance
- e2e: Through
codeaf do --jsonwith a stub cost ledger already beyondCODEAF_MAX_COST=0.001, assert prompt exit 3,"stop":"budget", answer texta dollar limit you set stopped it, and nodeadlinevalue. - e2e: Through the same door with a higher limit, assert the stub task leaves
pending, entersrunning, and the result is not stopped for budget. - Unit: A settlement with pending work, no running work, and an exhausted daily or run dollar allowance maps to budget immediately and cannot later map to deadline.
- The manual pages
internal/manual/chat/running-from-the-terminal.mdandinternal/manual/chat/models-and-cost.mdquote the budget-stop wording and JSON field, and the change entry'sinvalidatesnames the old belief that an unstartable dollar-limited task should wait for the time wall.
Filed from the 2026-09-21 staging verification of 17ae56d34; the lane reports, frames and e2e logs are in the owner's handoff folder for this promotion.
- Lingua principale
- Go
- Stelle
- 115
- Fork
- 14
- Merge medio
- 9h 35m
- PR unite (30g)
- 752
Preparare l'ambiente
Non abbiamo ancora controllato i file di configurazione di questo progetto. Parti dal suo README e consulta la nostra guida al primo contributo per i passaggi generali.
Come iniziare
- Leggi tutta la issue e poi la guida ai contributi del progetto.
- Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
- Fai un fork del repository e lavora su un branch.
- Apri una pull request che faccia riferimento al numero della issue.
Altre issue di Agent-Field/CodeAF
-
area:chat bug sev:papercut
Difficoltà 2/5 1-3 ore Idoneità per principianti 86/100
Agent-Field/CodeAF#1592 ·
I maintainer di solito rispondono entro 1 giorno
-
area:headless bug sev:critical
Difficoltà 2/5 1-3 ore Idoneità per principianti 88/100
Agent-Field/CodeAF#1566 ·
I maintainer di solito rispondono entro 1 giorno
-
area:chat bug sev:critical
Difficoltà 2/5 1-3 ore Idoneità per principianti 82/100
Agent-Field/CodeAF#1548 · 1 commento ·
I maintainer di solito rispondono entro 1 giorno
-
area:chat feature
Difficoltà 2/5 1-3 ore Idoneità per principianti 88/100
Agent-Field/CodeAF#1510 ·
I maintainer di solito rispondono entro 1 giorno
-
area:tests bug
Difficoltà 2/5 1-3 ore Idoneità per principianti 82/100
Agent-Field/CodeAF#1489 ·
I maintainer di solito rispondono entro 1 giorno
Tutte le issue di Agent-Field/CodeAF
Issue simili
-
area/proxy kind/bug priority/backlog triage/accepted
Difficoltà 2/5 1-3 ore Idoneità per principianti 68/100
lexfrei/cloudflare-tunnel-gateway-controller#840 ·
I maintainer di solito rispondono entro 1 giorno
-
kind/bug
Difficoltà 2/5 1-3 ore Idoneità per principianti 72/100
I maintainer di solito rispondono entro 7 giorni
-
bug needs triage
Difficoltà 2/5 1-3 ore Idoneità per principianti 78/100
I maintainer di solito rispondono entro 1 giorno
-
bug P2 reliability
Difficoltà 2/5 1-3 ore Idoneità per principianti 88/100
afreidah/s3-orchestrator#1564 ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 84/100
blinklabs-io/gouroboros#2577 ·
I maintainer di solito rispondono entro 1 giorno