Files
PaperClipAI/tests/runner-e2e/chat-catalog-domains/internal-tasks.ts
T
f4337bb5fb test(evals): land the chat catalog eval harness on master (#19614)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - The chat-feature evals measure whether agents complete real chat
requests through the product
> - The chat catalog eval harness and its repairs live on the
`chat-eval/master-candidate` branch, not on master
> - Today's work made 139 of 242 chat evals pass in verified single-case
live runs. Each product fix landed on master as its own pull request
> - The eval-side changes (setups, graders, fault seams, evidence
capture) still exist only on the candidate
> - This pull request lands the harness on master. Master stays the
source of truth for product behavior
> - The benefit is one tree: evals run against master commits, and
product fixes reach the evals as soon as they merge

## Linked Issues or Issue Description

Refs #19441, Refs #19599, Refs #20270, Refs #20334, Refs #20342, Refs
#20344, Refs #20354, Refs #20360, Refs #20373, Refs #20382, Refs #20409

**What existing behavior does this improve?**
The chat catalog eval harness and the product it evaluates live on a
long-running branch outside master. Eval fixes do not reach master, so
master-based eval runs use old setups and graders.

**Subsystem affected**
`tests/runner-e2e` chat catalog harness, plus a small set of product
changes that exist only on the candidate.

**Proposed change**
Land the harness on master. Every product fix from this eval work
already merged to master as its own pull request. Conflicts keep
master's product behavior.

## What Changed

- Most of the diff (808 of 851 files) is under `tests/runner-e2e/`:
chat-catalog domains, scenarios, descriptors, the TestOps manifest,
private evidence tooling, and their tests.
- Graders now check saved state, setup literals and rendered or clicked
links. Many keyword and regex checks on free-form replies are removed.
Each changed grader has a bumped `oracleVersion` and calibration tests
(a valid reply, a reworded valid reply, and broken variants that must
fail).
- The shared link helper `hasLink` no longer treats a list link (for
example `/COM/issues?view=mine`) as a link to every task without a URL
key.
- Setup fixes build real state through supported product routes: routine
history, agent grants through the board grant route, budget holds with a
real cap, and a storage-layer archive fault for
`inbox-search.archive-partial-failure`.
- Browser boundary journeys (sign-in, chat-disabled, onboarding) now
write the standard pass evidence files.
- Retargeted cases, approved by the maintainer:
`agent-chat.chat-disabled` (Conference Room gate),
`agents.restore-instructions` (no saved instruction history),
`access.manage-membership` (member permission edits stay board-only).
- The private evidence trusted files (`manual-private-handoff.ts`,
`manual-private-retention.ts`, `manual-private-report-assets.ts`) are
byte-identical to master.
- Product files outside `tests/` (43 files, +1,328 −49) carry
candidate-only changes that master does not yet have. Most are tests.
The main code changes are in `request-human-input.ts`,
`heartbeat-run-summary.ts`, `native-completion-feedback.ts`,
`native-run-finalizer.ts`, `openapi.ts`, `secrets.ts` and
`IssueChatLongThreadPerf.tsx`.

## Verification

- `pnpm vitest run chat-catalog-domains chat-catalog-contract.test.ts
chat-catalog-consistency.test.ts unit-launcher.test.mjs
--testTimeout=180000` in `tests/runner-e2e`: 269 files pass, 3,550 tests
pass.
- `node --test chat-full-catalog-testops-manifest.test.mjs`: 10 pass.
- `node tests/runner-e2e/source-sdk-public-inputs.mjs <HEAD> --check`:
no change.
- `tsc` for `server` and `tests/runner-e2e`: clean.
- Live single-case verification: 139 of 242 cases passed on paid runs,
and a person read the saved evidence for each. These are single-case
runs, not one full-cohort run.

## Risks

- Large diff. Most of it is test-only, but the 43 non-test files change
product code paths. Review those files first.
- Some graders are now less strict on reply wording. A person must read
replies for meaning on those cases.
- 42 cases need disposable outside accounts (Notion, Slack, GitHub,
Daytona and others) and stay blocked. `access.invite-teammate` stays
failing until local agent isolation lands, because the runner's durable
state keeps a live invite token.
- The paid workflow compares the trusted private-evidence files with
master. They are unchanged, so this risk is low.

## Model Used

Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use and code
execution, in Claude Code. Subagents used the same model.

## 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>
Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-11-10 08:02:11 -06:00

133 lines
9.3 KiB
TypeScript

import { savedDesignInstruction, withRenderedCommentReference, validRenderedCommentReference } from "./internal-comment-reference.js";
import { linkedQuestionTakeover, observeLinkedTaskQuestion } from "../chat-linked-question.js";
import { withEditOwnership } from "./internal-edit-ownership.js";
import { verifyDistinctTaskFixtures } from "./internal-distinct-fixtures.js";
import { delegatedTaskCases } from "./internal-task-creation.js";
import { taskPrivacyCases } from "./internal-privacy.js";
import { taskInventoryCases } from "./internal-task-inventory.js";
import { taskAttachmentCases } from "./internal-task-attachment.js";
import { taskCompletionCases } from "./internal-task-completion.js";
import { privateDelegationCases } from "./internal-private-delegation.js";
import type { ChatCatalogCase } from "../chat-catalog-contract.js";
import { project, task, goal, putDocument, entity, observation, checks, hasLink, sameTask, type Entity, type InternalObservation } from "./internal-support.js";
import { makeDefinition, retryDefinition, fixtureWorker } from "./internal-definition.js";
export function grade_tasks(value: unknown) {
const c = checks(); const o = observation(value);
c.add("complete-independent-observation", !!o, "Missing API, document, audit or rendered-link evidence cannot pass.");
if (!o) return c.result();
const before = o.state.task ? entity(o.state, "task") : undefined;
const current = before && o.tasks.find(row => row.id === before.id);
const savedProject = o.state.project ? entity(o.state, "project") : undefined;
const projectNow = savedProject && o.projects.find(row => row.id === savedProject.id);
const authored = (action: string, id: string) => o.activity.some(row => row.action === action && row.entityId === id && row.agentId === o.leadId);
const linkedTask = () => !!current && hasLink(o, "issues", current);
switch (o.scenarioId) {
case "tasks.edit-properties": {
const worker = entity(o.state, "worker");
c.add("exact-priority-assignee", !!current && current.priority === "high" && current.assigneeAgentId === worker.id, "The intended landing-page task has one eligible owner and High priority.");
c.add("no-unrequested-changes", !!before && sameTask(before, current, ["priority", "assigneeAgentId"]) && o.tasks.length === 1, "Task context and execution state remain intact; no workaround duplicate exists.");
c.add("saved-attribution", !!current && authored("issue.updated", current.id), "A real CEO-attributed mutation changed the requested task.");
c.add("task-link", linkedTask(), "The reply links to the edited task.");
break;
}
case "tasks.ambiguous-task": {
const other = entity(o.state, "otherTask");
const first = observation(o.state.clarification);
c.add("clarify-before-write", !!first && !!before && sameTask(before, first.tasks.find(row => row.id === before.id))
&& sameTask(other, first.tasks.find(row => row.id === other.id))
&& hasLink(first, "issues", before) && hasLink(first, "issues", other),
"The first completed turn changes neither task and gives the person links to both candidates.");
c.add("only-selected-task-changed", !!before && current?.priority === "high" && sameTask(before, current, ["priority"])
&& sameTask(other, o.tasks.find(row => row.id === other.id)) && o.tasks.length === 2,
"After the explicit Launch selection, only that task's requested priority changes.");
c.add("attributed-selection", !!current && authored("issue.updated", current.id)
&& !o.activity.some(row => row.entityId === other.id && row.agentId === o.leadId), "The selected mutation is attributed to the CEO and the other task has no agent changes.");
c.add("selected-result-link", linkedTask(), "The final conversation retains usable access to the exact changed task.");
break;
}
case "tasks.comment-with-reference": {
const design = entity(o.state, "design");
const added = o.comments.filter(row => row.authorAgentId === o.leadId && row.createdByRunId && o.runs.some(run => run.id === row.createdByRunId));
c.add("one-attributed-note", added.length === 1 && savedDesignInstruction(o, added[0]?.body ?? "", design) && validRenderedCommentReference(o, added[0]!, design), "One CEO-authored note retains a usable reference to the actual approved design task.");
c.add("records-preserved", !!before && sameTask(before, current) && sameTask(design, o.tasks.find(row => row.id === design.id)) && o.tasks.length === 2,
"The note neither changes either task nor creates another task.");
c.add("task-link", linkedTask(), "The original conversation links back to the task containing the note.");
break;
}
default: c.add("implemented-scenario", false, "No independent oracle is implemented for this scenario.");
}
return c.result();
}
const define = (id: string, setup: ChatCatalogCase["setup"]) => makeDefinition(id, setup, grade_tasks);
export const tasksCases: readonly ChatCatalogCase[] = [
...delegatedTaskCases,
...taskPrivacyCases,
...taskInventoryCases,
...taskAttachmentCases,
...taskCompletionCases,
...privateDelegationCases,
ambiguousTask(),
withEditOwnership(define("tasks.edit-properties", async ctx => {
const p = await project(ctx); const worker = await fixtureWorker(ctx);
const t = await task(ctx, "Landing page", p.id, { description: "Build the Launch landing page; this work is saved for later." });
return { project: p, worker, task: t };
})),
withRenderedCommentReference(define("tasks.comment-with-reference", async ctx => {
const p = await project(ctx); const design = await task(ctx, "Approved design", p.id, { status: "done", description: "Approved design: semantic HTML, clear contrast, one primary signup action." });
const t = await task(ctx, "Landing page", p.id, { description: "Build the launch website." });
return { project: p, design, task: t };
})),
];
function ambiguousTask(): ChatCatalogCase {
const base = define("tasks.ambiguous-task", async ctx => {
const p = await project(ctx); const otherProject = await project(ctx, "Experiments");
const t = await task(ctx, "Landing page", p.id);
// Public creation defaults to company-wide recent-title reuse, regardless of project.
const otherTask = await task(ctx, "Landing page", otherProject.id, { allowDuplicate: true });
await verifyDistinctTaskFixtures(ctx, [t, otherTask]);
return { project: p, task: t, otherProject, otherTask };
});
return { ...base, oracleVersion: "paperclip.chat-catalog.internal.ambiguous-task.v2", sourceFiles: [...base.sourceFiles, "chat-catalog-domains/internal-distinct-fixtures.ts", "chat-linked-question.ts"], expectedRunCount: 2, expectedHumanMessageCount: 2, attemptTimeoutMs: 420_000,
async act(ctx, state) {
const plain = await pollUntil({ label: "Task ambiguity clarification", deadlineAt: Date.now() + 150_000,
load: () => observeInternal(ctx, state), accept: o => o.runs.length === 1 && o.runs.every(row => row.status === "succeeded")
&& ctx.issue()?.conversationState === "waiting" });
const first = await observeLinkedTaskQuestion(ctx, state, plain);
state.clarification = { ...first, state: structuredClone(state) };
await ctx.input.evidence("chat-catalog-task-clarification.json", state.clarification);
const selected = entity(state, "task"); const other = entity(state, "otherTask");
if (!sameTask(selected, first.tasks.find(row => row.id === selected.id)) || !sameTask(other, first.tasks.find(row => row.id === other.id)))
throw new Error("Ambiguous task changed before the person chose a source");
const answer = `The Landing page task in Launch (${selected.identifier ?? selected.id}).`;
const card = first.linkedQuestionEvidence;
if (card) {
const fresh = (await observeLinkedTaskQuestion(ctx, state, plain)).linkedQuestionEvidence;
if (!fresh || fresh.interactionId !== card.interactionId || fresh.questionId !== card.questionId || fresh.payloadDigest !== card.payloadDigest)
throw new Error("The actual linked clarification question changed before the person's selection");
}
const answerRoot = card ? linkedQuestionTakeover(ctx.input.page, card) : ctx.input.page;
const textInput = answerRoot.getByTestId("question-text-answer-composer").last();
if (await textInput.isVisible()) {
await textInput.locator('[contenteditable="true"],textarea').first().fill(answer);
await answerRoot.getByRole("button", { name: /Submit answers|Submit|Continue/, exact: true }).last().click();
} else {
const otherChoice = answerRoot.getByRole("radio", { name: "Other", exact: true }).last();
if (await otherChoice.isVisible()) {
await otherChoice.click();
await answerRoot.getByTestId("question-other-answer-composer").last().locator('[contenteditable="true"],textarea').first().fill(answer);
await answerRoot.getByRole("button", { name: /Submit answers|Submit|Continue/, exact: true }).last().click();
} else {
if (card) throw new Error("The linked question has no supported explicit clarification input");
const { sendChatMessage } = await import("../chat-flow.js"); await sendChatMessage(ctx.input.page, answer);
}
}
},
};
}
import { observeInternal } from "./internal-support.js";
import { pollUntil } from "../api.js";