Files
PaperClipAI/tests/runner-e2e/lifecycle-baseline.ts
DottaandPaperclip 24a15c5afa test(evals): verify durable waiting across continuation checkpoints (#15564)
## Thinking Path

> - Paperclip manages agents and durable tasks across provider runs.
> - Human questions must preserve task ownership and stop work until a
real answer arrives.
> - The continuation eval checks the final work, but its lifecycle
oracle misses several broken waiting states.
> - A correct final answer can hide a stale execution lock or a lost
intermediate answer receipt.
> - This pull request checks each wait and retains every question and
run identity.
> - A source audit records which waiting operations belong to native
runtimes and which still require legacy API calls.

## Linked Issues or Issue Description

**What existing behavior does this improve?**

The existing Product E2E continuation oracle for human question and
approval journeys.

**Current behavior**

A pending interaction can pass the waiting check even when its task has
the wrong status, a stale lock, or a retry. Only the first pending
question and first run receipts are checked at the final checkpoint.

**Proposed behavior**

Require a healthy wait on the same task and assignee. Preserve all
intermediate run receipts and every pending question's answered
identity. Allow a paused native provider question only when its pending
runtime request identifies the running native run and execution lock.

Related: #15548 and #15554 cover earlier bookkeeping slices. #15544
changes production continuation summaries; this PR changes the eval
oracle and does not overlap that fix.

## What Changed

- Check waiting state, execution ownership, and all answer/run
identities in the existing lifecycle oracle.
- Add negative calibrations for broken records and positive coverage for
both semantic waits and paused provider questions.
- Retain task activity at each checkpoint for inspection of successful
persisted mutations.
- Verify 1,000-cent company and agent budget limits before continuation
work. Disable automatic cell rerolls.
- Document runtime ownership, existing coverage, and the remaining
instruction decision. Keep production instructions unchanged.

## Verification

- `pnpm test:e2e:runner:unit`: 1,867 Vitest tests pass, one skips; 128
Node tests pass.
- `pnpm test:e2e:runner:typecheck`: passes.
- `pnpm -r typecheck`: passes.
- `pnpm build`: passes.
- Negative calibration: 27 added cases fail against the prior oracle and
pass with these checks.
- Four existing local continuation cells are selected for a separate
bounded live canary. Live results are pending; no behavioral pass is
claimed here.
- The full local `pnpm test:run` suite was not repeated because embedded
PostgreSQL was unavailable in the preceding workspace verification.
Required Linux PR CI must pass before readiness.

## Risks

The stronger oracle can expose existing product or fixture defects. A
paused provider question and a terminal semantic wait have distinct
valid states. The four-cell canary does not qualify approval/review,
dependency unblock, crash races, remote execution, or general task
quality. Task activity records successful persisted writes, not failed
API attempts; repeated progress comments are not automatically defects.
Historical eval grades remain unchanged. No production scheduling,
prompt, tool, schema or migration changes.

## Model Used

OpenAI Codex, GPT-6-based, with tool use and code execution. The exact
serving model ID and context-window size are not exposed in this
session.

## 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 searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-10-08 08:45:36 -05:00

155 lines
6.1 KiB
TypeScript

import type { ContinuationCheckpoint } from "./continuation-scoring.js";
export interface LifecycleSnapshot {
executionRunId: string | null;
scheduledRetry: unknown;
activeRecoveryAction: unknown;
monitorNextCheckAt: string | null;
}
export type LifecycleCheckpoint = ContinuationCheckpoint & {
lifecycle?: LifecycleSnapshot;
// Successful persisted mutations; not a count of failed tool/HTTP attempts.
activity?: unknown[];
};
type Interaction = {
id?: string;
kind?: string;
status?: string;
sourceRunId?: string;
payload?: {
runtimeRequestId?: string;
target?: {
type?: string;
issueId?: string;
key?: string;
revisionId?: string;
};
};
};
const nonempty = (value: unknown): value is string =>
typeof value === "string" && value.trim().length > 0;
/** Semantic questions yield the run; provider-native questions instead pause
* inside it. Only a pending runtime request bound to that running native run
* permits retaining its execution lock at a waiting checkpoint. */
function hasHealthyWait(checkpoint: LifecycleCheckpoint, pending: Interaction[]) {
const state = checkpoint.lifecycle;
if (!state || state.scheduledRetry !== null || state.activeRecoveryAction !== null ||
state.monitorNextCheckAt !== null || checkpoint.runs.length === 0) return false;
const active = checkpoint.runs.filter(run => run.status !== "succeeded");
if (active.length === 0)
return checkpoint.issue.status === "in_review" && state.executionRunId === null;
if (active.length !== 1) return false;
const run = active[0];
return run.status === "running" && run.runtimeMode === "native" && nonempty(run.id) &&
state.executionRunId === run.id && ["in_progress", "in_review"].includes(checkpoint.issue.status) &&
pending.some(interaction => interaction.kind === "ask_user_questions" &&
nonempty(interaction.id) && interaction.sourceRunId === run.id &&
nonempty(interaction.payload?.runtimeRequestId));
}
/** Independent durable-state oracle. It never interprets response prose. */
export function gradeLifecycleBaseline(checkpoints: LifecycleCheckpoint[]) {
const checks: Array<{ id: string; passed: boolean; detail: string }> = [];
const check = (id: string, passed: boolean, detail: string) =>
checks.push({ id, passed, detail });
const final = checkpoints.find((c) => c.phase === "final");
const waiting = checkpoints.filter((c) => c.phase !== "final");
check(
"lifecycle.evidence-present",
!!final?.lifecycle &&
waiting.length > 0 &&
waiting.every((c) => !!c.lifecycle),
"Every checkpoint must retain lifecycle evidence from the public task API.",
);
const first = waiting[0];
check(
"lifecycle.task-owner-preserved",
nonempty(first?.issue.id) && nonempty(first?.issue.assigneeAgentId) &&
checkpoints.every(c => c.issue.id === first.issue.id &&
c.issue.assigneeAgentId === first.issue.assigneeAgentId),
"Waiting and resuming retain the same task and assigned agent.",
);
for (const c of waiting) {
const pending = (c.interactions as Interaction[]).filter(
(i) => i.status === "pending",
);
check(
`lifecycle.${c.phase}.durable-wait`,
pending.some(
(i) =>
nonempty(i.id) &&
[
"ask_user_questions",
"request_confirmation",
"request_approval",
].includes(i.kind ?? ""),
),
"Waiting must have an identifiable pending interaction, not only an assistant message.",
);
check(
`lifecycle.${c.phase}.wait-state`,
hasHealthyWait(c, pending),
"A settled wait is in_review with no execution lock, retry, recovery or monitor. A bound native provider question may keep its paused run and lock.",
);
for (const i of pending.filter(
(i) => i.payload?.target?.type === "issue_document",
)) {
const target = i.payload!.target!;
check(
`lifecycle.${c.phase}.revision:${i.id}`,
target.issueId === c.issue.id &&
c.documents.some(
(d) =>
d.key === target.key &&
typeof d.latestRevisionId === "string" &&
d.latestRevisionId === target.revisionId,
),
"Plan confirmation binds this task and the recorded current revision.",
);
}
}
check(
"lifecycle.final.no-active-path",
!!final?.lifecycle &&
final.issue.status === "done" &&
final.lifecycle.executionRunId === null &&
final.lifecycle.scheduledRetry === null &&
final.lifecycle.activeRecoveryAction === null &&
final.lifecycle.monitorNextCheckAt === null &&
final.runs.length > 0 &&
final.runs.every((r) => !["queued", "running"].includes(r.status)),
"Completed work has no live run, execution lock, scheduled retry, recovery or monitor.",
);
check(
"lifecycle.final.no-pending-interaction",
!!final &&
!(final.interactions as Interaction[]).some(
(i) => i.status === "pending",
),
"Completion has no unresolved interaction.",
);
if (first && final) {
const questions = new Map(waiting.flatMap(c =>
(c.interactions as Interaction[]).filter(i => i.kind === "ask_user_questions" && i.status === "pending")
.map(i => [i.id, i] as const)));
for (const question of questions.values()) {
const answers = (final.interactions as Interaction[]).filter(i => i.id === question.id);
check(
`lifecycle.answer:${question.id}`,
nonempty(question.id) && answers.length === 1 &&
answers[0].kind === "ask_user_questions" && answers[0].status === "answered",
"Every observed pending question retains exactly one answered identity, including later questions.",
);
}
const finalIds = new Set(final.runs.map(r => r.id));
check(
"lifecycle.final.preserved-runs",
checkpoints.every(c => c.runs.length > 0 && c.runs.every(r => nonempty(r.id)) &&
new Set(c.runs.map(r => r.id)).size === c.runs.length) &&
waiting.every(c => c.runs.every(r => finalIds.has(r.id))),
"All intermediate run receipts survive in the final snapshot without duplicated or empty IDs.",
);
}
return checks;
}