[codex] Retry max-turn exhausted heartbeats (#5096)

## Thinking Path

> - Paperclip orchestrates AI agents for autonomous companies, and
heartbeat execution is the control-plane loop that keeps assigned work
moving.
> - Max-turn exhaustion is a recoverable local-adapter stop condition
for Claude and Gemini agents when a run needs another heartbeat to
continue safely.
> - The previous behavior could leave max-turn continuation details hard
to inspect, and duplicate/stale continuation wakes could keep running
after issue state changed.
> - The adapter layer also needed to avoid trusting arbitrary
stdout/stderr text as scheduler control metadata.
> - This pull request adds bounded max-turn continuation scheduling,
visible retry state, structured stop metadata handling, and
stale/duplicate continuation guards.
> - The benefit is safer automatic continuation after max-turn stops,
clearer operator visibility, and fewer duplicate or stale agent runs.

## What Changed

- Replaces closed PR #4952, whose head repository was deleted.
- Rebases the recovered max-turn continuation branch onto current
`paperclipai/paperclip:master`.
- Adds max-turn continuation scheduling and retry-state plumbing for
heartbeat runs.
- Adds stale/duplicate continuation suppression when issue status,
ownership, or execution locks change.
- Normalizes Claude/Gemini max-turn detection around structured stop
metadata instead of unstructured stdout/stderr text.
- Surfaces max-turn continuation settings and retry visibility in the
board UI.
- Adds focused server, adapter, and UI tests for max-turn stop metadata,
retry scheduling, stale queued-run invalidation, adapter
parsing/execution, run ledger display, and agent config patching.

## Verification

- `pnpm install --no-frozen-lockfile` to refresh local dependencies
after rebasing onto current `master`.
- `pnpm run preflight:workspace-links && pnpm exec vitest run
server/src/__tests__/claude-local-adapter.test.ts
server/src/__tests__/claude-local-execute.test.ts
server/src/__tests__/gemini-local-adapter.test.ts
server/src/__tests__/gemini-local-execute.test.ts
server/src/__tests__/heartbeat-retry-scheduling.test.ts
server/src/__tests__/heartbeat-stale-queue-invalidation.test.ts
server/src/services/heartbeat-stop-metadata.test.ts
ui/src/components/IssueRunLedger.test.tsx
ui/src/lib/agent-config-patch.test.ts ui/src/lib/runRetryState.test.ts
--testTimeout=20000`
- `pnpm --filter @paperclipai/adapter-claude-local typecheck && pnpm
--filter @paperclipai/adapter-gemini-local typecheck && pnpm --filter
@paperclipai/server typecheck && pnpm --filter @paperclipai/ui
typecheck`
- UI screenshot note: the UI changes are limited to config/ledger state
rendering rather than layout changes; component/unit coverage above
verifies the rendered behavior.

## Risks

- Medium behavior risk: heartbeat retry gating now suppresses max-turn
continuations when issue state or execution locks drift, so any callers
that relied on stale continuations running will now see cancellation
instead.
- Low adapter risk: Claude/Gemini unstructured text no longer triggers
max-turn scheduler metadata, so only structured stop signals and Gemini
exit code 53 are trusted.
- No database migrations.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- OpenAI Codex coding agent, GPT-5-class model, tool-enabled local
repository editing and command execution.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] If this change affects the UI, I have included before/after
screenshots (not applicable: state/default rendering only; covered by
component/unit tests)
- [x] I have updated relevant documentation to reflect my changes (not
applicable: no user-facing command or docs contract changed)
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
This commit is contained in:
Dotta 2026-05-03 11:30:48 -05:00 committed by GitHub
parent 57229d0f24
commit 15eac43b43
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
21 changed files with 1915 additions and 120 deletions

View file

@ -27,7 +27,7 @@ import {
} from "@/components/ui/popover";
import { Button } from "@/components/ui/button";
import { FolderOpen, Heart, ChevronDown, X } from "lucide-react";
import { cn } from "../lib/utils";
import { asBoolean, asFiniteNumber, asObject, cn } from "../lib/utils";
import { extractModelName, extractProviderId } from "../lib/model-utils";
import { queryKeys } from "../lib/queryKeys";
import { useCompany } from "../context/CompanyContext";
@ -175,6 +175,19 @@ const claudeThinkingEffortOptions = [
{ id: "high", label: "High" },
] as const;
const MAX_TURN_CONTINUATION_DEFAULT_MAX_ATTEMPTS = 2;
const MAX_TURN_CONTINUATION_MAX_ATTEMPTS_CAP = 10;
const MAX_TURN_CONTINUATION_DEFAULT_DELAY_SEC = 1;
const MAX_TURN_CONTINUATION_MAX_DELAY_SEC = 300;
function clampInteger(value: number, min: number, max: number) {
return Math.max(min, Math.min(max, Math.floor(value)));
}
function clampDelayMsFromSeconds(value: number) {
return clampInteger(value, 0, MAX_TURN_CONTINUATION_MAX_DELAY_SEC) * 1000;
}
/* ---- Form ---- */
@ -628,6 +641,27 @@ export function AgentConfigForm(props: AgentConfigFormProps) {
const currentDefaultEnvironmentId = isCreate
? val!.defaultEnvironmentId ?? ""
: eff("identity", "defaultEnvironmentId", props.agent.defaultEnvironmentId ?? "");
const effectiveHeartbeat = asObject(effectiveRuntimeConfig.heartbeat);
const maxTurnContinuation = asObject(effectiveHeartbeat.maxTurnContinuation);
const maxTurnContinuationEnabled = asBoolean(maxTurnContinuation.enabled, true);
const maxTurnContinuationMaxAttempts = clampInteger(
asFiniteNumber(maxTurnContinuation.maxAttempts, MAX_TURN_CONTINUATION_DEFAULT_MAX_ATTEMPTS),
0,
MAX_TURN_CONTINUATION_MAX_ATTEMPTS_CAP,
);
const maxTurnContinuationDelaySec = clampInteger(
asFiniteNumber(maxTurnContinuation.delayMs, MAX_TURN_CONTINUATION_DEFAULT_DELAY_SEC * 1000) / 1000,
0,
MAX_TURN_CONTINUATION_MAX_DELAY_SEC,
);
function updateMaxTurnContinuation(patch: Record<string, unknown>) {
mark("heartbeat", "maxTurnContinuation", {
...maxTurnContinuation,
...patch,
});
}
return (
<div className={cn("relative", cards && "space-y-6")}>
{/* ---- Floating Save button (edit mode, when dirty) ---- */}
@ -1182,6 +1216,40 @@ export function AgentConfigForm(props: AgentConfigFormProps) {
className={inputClass}
/>
</Field>
<div className="rounded-md border border-border/70 px-3 py-2">
<ToggleField
label="Continue after max-turn stop"
hint={help.maxTurnContinuationEnabled}
checked={maxTurnContinuationEnabled}
onChange={(v) => updateMaxTurnContinuation({ enabled: v })}
/>
{maxTurnContinuationEnabled ? (
<div className="mt-3 grid gap-3 sm:grid-cols-2">
<Field label="Continuation attempts" hint={help.maxTurnContinuationMaxAttempts}>
<DraftNumberInput
value={maxTurnContinuationMaxAttempts}
onCommit={(v) =>
updateMaxTurnContinuation({
maxAttempts: clampInteger(v, 0, MAX_TURN_CONTINUATION_MAX_ATTEMPTS_CAP),
})}
immediate
className={inputClass}
/>
</Field>
<Field label="Continuation delay (sec)" hint={help.maxTurnContinuationDelaySec}>
<DraftNumberInput
value={maxTurnContinuationDelaySec}
onCommit={(v) =>
updateMaxTurnContinuation({
delayMs: clampDelayMsFromSeconds(v),
})}
immediate
className={inputClass}
/>
</Field>
</div>
) : null}
</div>
</div>
</CollapsibleSection>
</div>

View file

@ -317,6 +317,44 @@ describe("IssueRunLedger", () => {
expect(container.textContent).toContain("Manual intervention required");
});
it("labels max-turn stops and continuation retries without confusing them with per-run turns", () => {
renderLedger({
runs: [
createRun({
runId: "run-scheduled-continuation",
status: "scheduled_retry",
finishedAt: null,
livenessState: null,
livenessReason: null,
retryOfRunId: "run-max-turns",
scheduledRetryAt: "2026-04-18T20:15:00.000Z",
scheduledRetryAttempt: 1,
scheduledRetryReason: "max_turns_continuation",
}),
createRun({
runId: "run-max-turns",
resultJson: { stopReason: "max_turns_exhausted" },
createdAt: "2026-04-18T19:57:00.000Z",
}),
createRun({
runId: "run-continuation-exhausted",
status: "failed",
createdAt: "2026-04-18T19:56:00.000Z",
retryOfRunId: "run-max-turns",
scheduledRetryAttempt: 3,
scheduledRetryReason: "max_turns_continuation",
retryExhaustedReason: "Bounded retry exhausted after 3 scheduled attempts; no further automatic retry will be queued",
}),
],
});
expect(container.textContent).toContain("Continuation scheduled");
expect(container.textContent).toContain("Max-turn continuation");
expect(container.textContent).toContain("Next continuation");
expect(container.textContent).toContain("Stop max turns exhausted");
expect(container.textContent).toContain("Continuation exhausted");
});
it("shows timeout, cancel, and budget stop reasons without raw logs", () => {
renderLedger({
runs: [

View file

@ -311,6 +311,7 @@ function stopReasonLabel(run: RunForIssue) {
if (timeoutFired || stopReason === "timeout") {
return timeoutText ? `timeout (${timeoutText})` : "timeout";
}
if (stopReason === "max_turns_exhausted" || stopReason === "turn_limit_exhausted") return "max turns exhausted";
if (stopReason === "budget_paused") return "budget paused";
if (stopReason === "cancelled") return "cancelled";
if (stopReason === "paused") return "paused by board";

View file

@ -56,6 +56,9 @@ export const help: Record<string, string> = {
wakeOnDemand: "Allow this agent to be woken by assignments, API calls, UI actions, or automated systems.",
cooldownSec: "Minimum seconds between consecutive heartbeat runs.",
maxConcurrentRuns: "Maximum number of heartbeat runs that can execute simultaneously for this agent.",
maxTurnContinuationEnabled: "Automatically queue bounded continuation runs when an adapter stops because its per-run turn cap was exhausted.",
maxTurnContinuationMaxAttempts: "Maximum automatic continuations after one max-turn stop. This is separate from max turns per run.",
maxTurnContinuationDelaySec: "Seconds to wait before starting each max-turn continuation.",
budgetMonthlyCents: "Monthly spending limit in cents. 0 means no limit.",
};

View file

@ -108,6 +108,35 @@ describe("buildAgentUpdatePatch", () => {
expect(patch.adapterConfig).toBeUndefined();
});
it("writes max-turn continuation policy under runtimeConfig.heartbeat", () => {
const patch = buildAgentUpdatePatch(
makeAgent(),
makeOverlay({
heartbeat: {
maxTurnContinuation: {
enabled: true,
maxAttempts: 3,
delayMs: 1000,
},
},
}),
);
expect(patch).toEqual({
runtimeConfig: {
heartbeat: {
enabled: true,
intervalSec: 300,
maxTurnContinuation: {
enabled: true,
maxAttempts: 3,
delayMs: 1000,
},
},
},
});
});
it("merges cheap profile changes onto existing runtimeConfig.modelProfiles state", () => {
const agent = makeAgent();
agent.runtimeConfig = {

View file

@ -5,6 +5,7 @@ describe("runRetryState", () => {
it("formats internal retry reasons for operators", () => {
expect(formatRetryReason("transient_failure")).toBe("Transient failure");
expect(formatRetryReason("issue_continuation_needed")).toBe("Continuation needed");
expect(formatRetryReason("max_turns_continuation")).toBe("Max-turn continuation");
expect(formatRetryReason("custom_reason")).toBe("custom reason");
});
@ -24,6 +25,22 @@ describe("runRetryState", () => {
});
});
it("describes max-turn continuation retries distinctly", () => {
expect(
describeRunRetryState({
status: "scheduled_retry",
retryOfRunId: "run-max-turns",
scheduledRetryAttempt: 1,
scheduledRetryReason: "max_turns_continuation",
scheduledRetryAt: "2026-04-18T20:15:00.000Z",
}),
).toMatchObject({
kind: "scheduled",
badgeLabel: "Continuation scheduled",
detail: "Attempt 1 · Max-turn continuation",
});
});
it("describes exhausted retries", () => {
expect(
describeRunRetryState({

View file

@ -24,6 +24,7 @@ const RETRY_REASON_LABELS: Record<string, string> = {
process_lost: "Process lost",
assignment_recovery: "Assignment recovery",
issue_continuation_needed: "Continuation needed",
max_turns_continuation: "Max-turn continuation",
};
function readNonEmptyString(value: unknown) {
@ -51,6 +52,7 @@ export function describeRunRetryState(run: RetryAwareRun): RunRetryStateSummary
const retryOfRunId = readNonEmptyString(run.retryOfRunId);
const exhaustedReason = readNonEmptyString(run.retryExhaustedReason);
const dueAt = run.scheduledRetryAt ? formatDateTime(run.scheduledRetryAt) : null;
const isMaxTurnContinuation = run.scheduledRetryReason === "max_turns_continuation";
const hasRetryMetadata =
Boolean(retryOfRunId)
|| Boolean(reasonLabel)
@ -63,10 +65,12 @@ export function describeRunRetryState(run: RetryAwareRun): RunRetryStateSummary
if (run.status === "scheduled_retry") {
return {
kind: "scheduled",
badgeLabel: "Retry scheduled",
badgeLabel: isMaxTurnContinuation ? "Continuation scheduled" : "Retry scheduled",
tone: "border-cyan-500/30 bg-cyan-500/10 text-cyan-700 dark:text-cyan-300",
detail: joinFragments([attemptLabel, reasonLabel]),
secondary: dueAt ? `Next retry ${dueAt}` : "Next retry pending schedule",
secondary: dueAt
? `${isMaxTurnContinuation ? "Next continuation" : "Next retry"} ${dueAt}`
: `${isMaxTurnContinuation ? "Next continuation" : "Next retry"} pending schedule`,
retryOfRunId,
};
}
@ -74,7 +78,7 @@ export function describeRunRetryState(run: RetryAwareRun): RunRetryStateSummary
if (exhaustedReason) {
return {
kind: "exhausted",
badgeLabel: "Retry exhausted",
badgeLabel: isMaxTurnContinuation ? "Continuation exhausted" : "Retry exhausted",
tone: "border-amber-500/30 bg-amber-500/10 text-amber-700 dark:text-amber-300",
detail: joinFragments([attemptLabel, reasonLabel, "Automatic retries exhausted"]),
secondary: exhaustedReason.includes("Manual intervention required")
@ -86,7 +90,7 @@ export function describeRunRetryState(run: RetryAwareRun): RunRetryStateSummary
return {
kind: "attempted",
badgeLabel: "Retried run",
badgeLabel: isMaxTurnContinuation ? "Continued run" : "Retried run",
tone: "border-slate-500/20 bg-slate-500/10 text-slate-700 dark:text-slate-300",
detail: joinFragments([attemptLabel, reasonLabel]),
secondary: null,

View file

@ -7,6 +7,20 @@ export function cn(...inputs: ClassValue[]) {
return twMerge(clsx(inputs));
}
export function asObject(value: unknown): Record<string, unknown> {
return typeof value === "object" && value !== null && !Array.isArray(value)
? value as Record<string, unknown>
: {};
}
export function asBoolean(value: unknown, fallback: boolean) {
return typeof value === "boolean" ? value : fallback;
}
export function asFiniteNumber(value: unknown, fallback: number) {
return typeof value === "number" && Number.isFinite(value) ? value : fallback;
}
export function formatCents(cents: number): string {
return `$${(cents / 100).toLocaleString("en-US", { minimumFractionDigits: 2, maximumFractionDigits: 2 })}`;
}