Files
PaperClipAI/tests/runner-e2e/chat-catalog-run-policy.ts
T
DottaandClaude Opus 5.5 cbd636d652 fix(evals): wait for rendered replies, bound runs by role and grade chat outcomes structurally (#20685)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - The chat-feature eval catalog runs real chat journeys and grades
saved state, rendered links and run ledgers.
> - The second full-cohort run on master `9296c74e8d` failed 15 cases
where the product did the right thing and the harness did not.
> - The causes were three harness classes: links read before the reply
rendered, raw run counts that rejected product-caused follow-up runs,
and keyword graders on free-form replies. Four browser locators also
targeted retired controls.
> - This pull request fixes those harness layers. It does not change the
product.
> - The benefit is that these cases measure the feature again: correct
behaviour passes, and the relevant broken behaviour still fails.

## Linked Issues or Issue Description

**What happened**
In the second full-cohort chat-eval run on master `9296c74e8d`, these
cases failed although the reply and the saved state were correct:
`agents.choose-primary`, `cases.case-history`,
`budgets-costs.set-budget` (links read about 2s before the reply
rendered); `task-coordination.independent-review`,
`cases.update-case-facts`, `routines.manage-other-owner`,
`questions-and-approvals.review-plan` (a legitimate continuation or the
product's completion follow-up exceeded a raw run count or was
cancelled); `cases.cases-disabled`, `inbox-search.find-design`,
`navigation-preferences.broken-link`, `cases.browse-cases` (keyword or
clause graders rejected correct replies); `agent-chat.conference-room`,
`workspaces.inspect-git`, `task-coordination.block-on-design`,
`skills.author-skill` (browser steps used stale settings or controls).

**Expected behavior**
A correct reply and correct saved state pass. A wrong outcome, extra or
duplicate agent work, a substitution or a missing route still fails.

**Steps to reproduce**
Run `runner-full-stack-e2e.yml` with `cadence=manual`, `all=false` and
an `id` list of the cases above against master `9296c74e8d`. Read
`result.json` and `snapshots/` in each cell's evidence.

Refs #20600

## What Changed

- `chat-rendered-links.ts`: `collectConversationLinks` waits (at most
10s, once per reply per page) for the latest saved lead reply to render
when its conversation is open. A page that does not show the
conversation does not wait. `observeInternal` and the flow conversation
read use it.
- `budgets-costs.set-budget`: waits for the rendered reply before it
reads anchors. It fills the budget only after the form shows the saved
cap, and checks that the typed value stays.
- New `chat-catalog-domains/run-roles.ts`: identifies a run by role (a
worker woken by a comment that a named run saved on the task; the CEO's
chat completion follow-up).
- `task-coordination.independent-review`: admits one Morgan continuation
woken by the CEO's authorization comment and one Riley review run. The
review request may come from that continuation's native status decision.
Other runs still fail.
- `cases.update-case-facts`: a phase settles when its CEO turn and at
most one Morgan continuation woken by that turn's task comment finish. A
new `role-bound-runs` check rejects any other run. `terminal-counts`
gets an optional `maximumTerminalStatusCounts`.
- `routines.manage-other-owner`: admits one Morgan run on a task that
the CEO created for Morgan. A schedule change must be Morgan's own write
from that run. The ownership check is structural (a linked owner task,
or a link to the routine or Morgan's page). The keyword checks on the
reply are removed.
- `questions-and-approvals.review-plan`: after the stale-review check,
lets the product's completion follow-up settle before final run custody.
The existing grader already validates it.
- `cases.cases-disabled`: the negation-blind false-claim regex is
removed. The case passes with a reply link to the Experimental settings
page (the scripted person follows it and sees the Cases control) or a
pending choice card. Saved state still proves that nothing was created.
- `inbox-search.find-design`: the per-clause regex is removed. The first
task link in the reply must be the task with the approved revision.
- `navigation-preferences.broken-link`: a link to the similar task fails
only when its label names the original (identifier or exact title). A
separate mention passes.
- `cases.browse-cases`: an excluded case must not be a result row. When
the reply has no result list, every case link is a result, so an
excluded case must not be linked.
- Browser steps: `conference-room` reloads after the fixture enables the
flag. `inspect-git` and `block-on-design` open the panel with "Show
properties" and pick the tab from the side panel launcher.
`author-skill` does not repeat the full inspection in `verifyResult`
when `act` already inspected the same saved version.
- Each changed case has a new `oracleVersion`.
`chat-catalog-descriptors.json` is in sync.

## Verification

- `pnpm exec vitest run --config tests/runner-e2e/vitest.config.ts <119
affected test files> --testTimeout=180000`: 1673 passed, 3 skipped. One
file (`chat-factual-query-input.test.ts`) did not load in the local
worktree because `@paperclipai/plugin-sdk` is not built there. It does
not import changed code. CI runs it.
- `chat-catalog-consistency.test.ts` passes (242 descriptors).
- `npx tsc --noEmit -p tests/runner-e2e`: no errors in changed files.
- `node tests/runner-e2e/source-sdk-public-inputs.mjs <HEAD> --check`:
no pin change.
- `node --test` on the three `chat-full-catalog-testops-*.test.mjs`
files: 40 passed.
- Calibration against retained evidence: the retained replies and run
shapes of cases-disabled, browse-cases, find-design, broken-link,
independent-review, update-case-facts and manage-other-owner now pass.
Broken variants fail (closed case as a result row, earlier design linked
first, original label on the similar task's link, unprompted or
duplicate worker runs, a second review run, a routine change outside the
owner's run, no route link).
- Paid runs on this branch: see the PR comments.

## Risks

- Low risk. Only eval harness code changes.
- The structural graders do not read reply wording. A reply that words
the outcome wrongly but links the right records can pass. A person reads
the reply.
- `navigation-preferences.sidebar-order` is not changed. The shipped
sidebar has no per-project order or collapsible Agents section, so the
case premise needs an owner decision.

> 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

- Anthropic Claude Opus 5.5 (`claude-opus-5-5`), tool use and code
execution in Claude Code.

## 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

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-10 20:24:41 -06:00

113 lines
9.0 KiB
TypeScript

import type { ChatCatalogCase, ChatCatalogGrade } from "./chat-catalog-contract.js";
import type { ChatIssue, ChatRun } from "./chat-flow.js";
type CancellationCase = Pick<ChatCatalogCase, "runPolicy"> & Partial<Pick<ChatCatalogCase, "scenarioId" | "expectedCancelledWorkerRun">>;
export function assertExpectedWorkerCancellation(implementation: CancellationCase) {
const flag = implementation.runPolicy?.kind === "source-settled-with-active-work"
? implementation.runPolicy.expectedWorkerCancellation : undefined;
if (flag === undefined ? implementation.expectedCancelledWorkerRun !== undefined
: flag !== "owned-task-reassignment" || implementation.scenarioId !== "tasks.edit-properties" || typeof implementation.expectedCancelledWorkerRun !== "function")
throw new Error("Missing or undeclared exact task-reassignment cancellation proof");
}
function permitsReassignment(implementation: CancellationCase) {
return implementation.scenarioId === "tasks.edit-properties" && implementation.runPolicy?.kind === "source-settled-with-active-work"
&& implementation.runPolicy.expectedWorkerCancellation === "owned-task-reassignment" && typeof implementation.expectedCancelledWorkerRun === "function";
}
export function chatCatalogAllowedRunStatuses(implementation: CancellationCase): string[] {
const policy = implementation.runPolicy;
return policy?.kind === "terminal-counts" ? Object.keys(policy.terminalStatusCounts)
: policy?.kind === "source-settled-with-active-work" ? ["queued", "running", "succeeded", ...(permitsReassignment(implementation) ? ["cancelled"] : [])] : ["succeeded"];
}
const taskId = (run: ChatRun) => (run as ChatRun & { nativeIssueId?: string }).nativeIssueId ?? run.contextSnapshot?.issueId;
const record = (value: unknown) => value && typeof value === "object" ? value as Record<string, unknown> : {};
/** Early polling can defer only the saved, independently held worker, never a source or unrelated cancellation. */
export function chatCatalogRunStatusAllowed(input: {
implementation: CancellationCase; run: ChatRun; companyId: string; leadId: string; source?: ChatIssue; observation?: unknown;
}): boolean {
const { implementation, run } = input;
if (run.status !== "cancelled" || !permitsReassignment(implementation)) return chatCatalogAllowedRunStatuses(implementation).includes(run.status);
const state = record(record(input.observation).state), held = record(state.heldRun), task = record(state.heldTask), worker = record(state.riley);
return held.id === run.id && held.status === "running" && held.companyId === input.companyId && held.agentId === worker.id
&& run.companyId === input.companyId && run.agentId === worker.id && run.agentId !== input.leadId
&& task.companyId === input.companyId && task.assigneeAgentId === worker.id && task.executionRunId === run.id
&& task.id === taskId(run) && task.id !== input.source?.id && run.errorCode === "issue_reassigned";
}
/**
* A queued run that never started and that the server cancelled because its own
* non-source task reached a terminal status is a task lifecycle event, not a
* provider failure. Early polling must not stop on it; the run stays in the
* ledger, and the run-bound and case checks still grade it.
*/
export function chatCatalogNeverStartedTerminalTaskRun(run: ChatRun, source?: ChatIssue): boolean {
const issueId = taskId(run);
return run.status === "cancelled" && run.errorCode === "issue_terminal_status" && !run.startedAt
&& record(run.resultJson).stopReason === "issue_terminal_status"
&& typeof issueId === "string" && issueId.length > 0 && issueId !== source?.id;
}
export function gradeChatCatalogRuns(input: {
implementation: Pick<ChatCatalogCase, "expectedRunCount" | "minimumRunCount" | "runPolicy"> & CancellationCase;
companyId: string; leadId: string; source?: ChatIssue; runs: ChatRun[];
observation?: unknown;
}): ChatCatalogGrade {
const { implementation: c, runs, source } = input;
const checks = [
{ id: "catalog-bounded-native-runs", passed: runs.length >= (c.minimumRunCount ?? c.expectedRunCount) && runs.length <= c.expectedRunCount
&& runs.every(run => run.companyId === input.companyId && run.runtimeMode === "native" && chatCatalogAllowedRunStatuses(c).includes(run.status)),
detail: "Observed provider runs satisfy declared count, company, native runtime and allowed outcome bounds." },
];
if (c.expectedCancelledWorkerRun !== undefined || c.runPolicy?.kind === "source-settled-with-active-work" && c.runPolicy.expectedWorkerCancellation !== undefined) {
let valid = true; try { assertExpectedWorkerCancellation(c); } catch { valid = false; }
checks.push({ id: "catalog-worker-cancellation-declaration", passed: valid,
detail: "Only edit-properties may declare the exact independently proved worker reassignment cancellation." });
}
if (c.runPolicy?.kind === "terminal-counts") {
const policy = c.runPolicy, maximum = policy.maximumTerminalStatusCounts ?? {};
checks.push({ id: "catalog-exact-run-outcomes", passed: Object.entries(policy.terminalStatusCounts).every(([status, count]) => {
const actual = runs.filter(run => run.status === status).length, upper = maximum[status as keyof typeof maximum] ?? count;
return actual >= count! && actual <= Math.max(count!, upper);
}) && Object.keys(maximum).every(status => status in policy.terminalStatusCounts),
detail: "The independently observed ledger matches every declared successful/cancelled/failed terminal outcome count (or its declared role-bound range)." });
} else if (c.runPolicy?.kind === "source-settled-with-active-work") {
const sourceRuns = runs.filter(run => taskId(run) === source?.id);
const workers = runs.filter(run => taskId(run) !== source?.id);
const cancelled = runs.filter(run => run.status === "cancelled");
const declared = permitsReassignment(c);
let expected: ReturnType<NonNullable<ChatCatalogCase["expectedCancelledWorkerRun"]>>;
try { expected = declared ? c.expectedCancelledWorkerRun!(input.observation) : undefined; } catch { expected = undefined; }
const actual = expected && workers.find(run => run.id === expected.runId);
const acknowledgement = actual?.resultJson?.nativeCancellation as Record<string, unknown> | undefined;
const exact = !!expected && cancelled.length === 1 && cancelled[0]?.id === expected.runId && !!actual
&& chatCatalogRunStatusAllowed({ ...input, run: actual })
&& actual.companyId === input.companyId && actual.agentId === expected.workerId && actual.agentId !== input.leadId
&& taskId(actual) === expected.taskId && expected.taskId !== source?.id && actual.errorCode === "issue_reassigned"
&& actual.resultJson?.reassignmentStopConfirmed === true
&& acknowledgement?.schema === "paperclip.native-cancellation.v1" && acknowledgement.runId === actual.id
&& acknowledgement.issueId === expected.taskId && acknowledgement.companyId === input.companyId
&& acknowledgement.dispatchState === "acknowledged" && typeof acknowledgement.acknowledgedAt === "string" && Number.isFinite(Date.parse(acknowledgement.acknowledgedAt))
&& typeof acknowledgement.intentAuditId === "string" && acknowledgement.intentAuditId.length > 0
&& typeof acknowledgement.acknowledgementAuditId === "string" && acknowledgement.acknowledgementAuditId.length > 0
&& sourceRuns.some(run => run.id === expected.sourceRunId && run.agentId === input.leadId && run.status === "succeeded");
checks.push({ id: "catalog-exact-owned-worker-cancellation", passed: cancelled.length === 0 ? expected === undefined : declared && exact,
detail: "Only the exact independently proved task-reassignment cancellation is accepted; final actor, chronology and saved-state proof remains mandatory in the case callback." });
checks.push({ id: "catalog-settled-source-separate-work", passed: !!source && source.companyId === input.companyId
&& source.conversationState === "waiting" && sourceRuns.length > 0 && sourceRuns.every(run => run.agentId === input.leadId && run.status === "succeeded")
&& workers.every(run => typeof taskId(run) === "string" && run.agentId !== input.leadId)
&& workers.filter(run => ["queued", "running"].includes(run.status)).length <= c.runPolicy.maximumActiveWorkerRuns,
detail: "The CEO source is settled; unfinished worker runs are separately task-bound and bounded. The case oracle must additionally bind their exact task and assignee." });
}
return { passed: checks.every(check => check.passed), checks };
}
export type ChatCatalogFinalRunVerifier = (source: ChatIssue | undefined, runs: ChatRun[]) => ChatCatalogGrade;
/** Preserve the loaded case's proof callback for a fresh outer ledger read. */
export function createChatCatalogFinalRunVerifier(input: {
implementation: Parameters<typeof gradeChatCatalogRuns>[0]["implementation"];
companyId: string;
leadId: string;
observation: unknown;
}): ChatCatalogFinalRunVerifier {
assertExpectedWorkerCancellation(input.implementation);
return (source, runs) => gradeChatCatalogRuns({ ...input, source, runs });
}