Files
PaperClipAI/tests/runner-e2e/chat-catalog-domains/runtime-history.test.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

57 lines
9.8 KiB
TypeScript

import { it, expect } from "vitest";
import { gradeRememberPreview, gradeBrowserReconnect } from "./runtime-history.js";
function preview() {
const artifact = { contentVerified: true, attachmentId: "blue", sha256: "hash", content: "<h1>Blue Selected Orchid</h1>", workProductId: "product" };
return { scenarioId: "agent-chat.remember-and-follow-up", origin: "http://fixture.invalid", companyId: "company", leadId: "ceo",
state: { previewBefore: artifact, previewAfter: { ...artifact }, designMarker: "Selected Orchid", filename: "blue-preview.html", reloaded: true, historyRetained: true,
choiceEstablished: true, canonicalRunIds: ["query"], queryReply: "Here is the blue preview we selected.", queryLinks: ["/api/attachments/blue/content?download=1"],
linkedPreviews: [{ attachmentId: "blue", issueId: "source", originalFilename: "blue-preview.html", sha256: "hash", contentVerified: true }],
unknownDesignPhase: { attributed: true, answeredTurn: true, runIds: ["unknown"], reply: "Which green design?", questionIds: [] as string[], createdArtifacts: [] as string[], createdRevisions: [] as string[] } },
agents: [], tasks: [] as Array<{ id: string; description?: string }>, links: ["/api/attachments/blue/content?download=1"], activity: [], reply: "Blue preview", humanQueries: [],
runs: [{ id: "prior", agentId: "ceo", status: "succeeded" }, { id: "query", agentId: "ceo", status: "succeeded" }, { id: "unknown", agentId: "ceo", status: "succeeded" }] };
}
it("accepts the same real blue preview after reload", () => expect(gradeRememberPreview(preview()).passed).toBe(true));
it("does not grade reply wording in either phase", () => { const o = preview(); o.state.queryReply = "Sure, opening it again."; o.state.unknownDesignPhase.reply = "Hmm, I only have one design on record."; expect(gradeRememberPreview(o).passed).toBe(true); });
it("accepts a byte-identical re-registration of the selected preview", () => { const o = preview(); o.state.queryLinks = ["/api/attachments/copy/content?download=1"]; o.state.linkedPreviews = [{ ...o.state.linkedPreviews[0]!, attachmentId: "copy" }]; expect(gradeRememberPreview(o).passed).toBe(true); });
it("accepts a clarification asked as a pending question card", () => { const o = preview(); Object.assign(o.state.unknownDesignPhase, { attributed: false, reply: "", questionIds: ["card"] }); expect(gradeRememberPreview(o).passed).toBe(true); });
it("rejects absent history evidence", () => expect(gradeRememberPreview(undefined).passed).toBe(false));
it("rejects an inaccessible or substituted preview", () => { const o = preview(); o.state.previewAfter.attachmentId = "other"; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects linking a regenerated preview with different bytes", () => { const o = preview(); o.state.linkedPreviews = [{ ...o.state.linkedPreviews[0]!, sha256: "regenerated" }]; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects a linked preview whose download does not match its record", () => { const o = preview(); o.state.linkedPreviews = [{ ...o.state.linkedPreviews[0]!, contentVerified: false }]; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects a preview link only present in an earlier phase", () => { const o = preview(); o.state.queryLinks = []; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects asking which design instead of returning the preview", () => { const o = preview(); o.state.queryReply = "Which design did you mean?"; o.state.queryLinks = []; o.state.linkedPreviews = []; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects inventing blue agreement without established history", () => { const o = preview(); o.state.choiceEstablished = false; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects unrequested follow-up work", () => { const o = preview(); o.tasks = [{ id: "followup", description: "Use the blue design" }]; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects fabricating the unestablished alternate preview", () => { const o = preview(); o.state.unknownDesignPhase.createdArtifacts = ["green-preview"]; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects changing saved documents before the clarification is answered", () => { const o = preview(); o.state.unknownDesignPhase.createdRevisions = ["green-selection"]; expect(gradeRememberPreview(o).passed).toBe(false); });
it("rejects clarification from the wrong execution phase", () => { const o = preview(); o.state.unknownDesignPhase.answeredTurn = false; expect(gradeRememberPreview(o).passed).toBe(false); });
function reconnect() {
return { scenarioId: "recovery.browser-reconnect", origin: "http://fixture.invalid", companyId: "company", leadId: "ceo",
state: { task: { id: "website", identifier: "ACORN-1" }, worker: { id: "morgan" }, workerRunId: "worker",
workerWasReallyWaiting: true, offlineObserved: true, workerStatusBeforeDisconnect: "running", workerStatusWhileOffline: "succeeded",
taskStatusWhileOffline: "done", serverAdvancedWhileOffline: true, reconnected: true, historyRetained: true,
offlineCommentsRetained: true, queryReply: "The website ACORN-1 finished while you were away.", canonicalRunIds: ["query"],
remainingPhase: { attributed: true, reply: "That follow-up is still running and waiting for the brief.", taskId: "followup", runId: "remaining", offlineObserved: true, runningWhileOffline: true, reconnected: true, runStatusAtReply: "running", taskStatusAtReply: "in_progress" },
failedReconnectPhase: { attributed: true, reply: "Riley's attempt failed with a provider model configuration error.", runId: "real-failure", taskId: "failed-task", offlineObserved: true, failedWhileOffline: true, reconnected: true } },
agents: [], tasks: [{ id: "website", status: "done" }, { id: "followup", status: "done" }, { id: "failed-task", status: "blocked" }], documents: [{ id: "delivery", title: "Website delivery", latestRevisionId: "saved" }],
revisions: [{ id: "saved", documentId: "delivery", createdByRunId: "worker", createdByAgentId: "morgan" }],
links: [], activity: [], reply: "", humanQueries: ["Track this", "Did you finish while I was disconnected?", "Is the follow-up finished?", "Did Riley finish?"],
runs: [{ id: "worker", agentId: "morgan", status: "succeeded" }, { id: "prior", agentId: "ceo", status: "succeeded" }, { id: "query", agentId: "ceo", status: "succeeded" }, { id: "remaining", agentId: "morgan", status: "succeeded" }, { id: "status", agentId: "ceo", status: "succeeded" },
{ id: "real-failure", agentId: "riley", status: "failed", runtimeMode: "native", errorCode: "native_provider_model_rejected" }, { id: "failed-status", agentId: "ceo", status: "succeeded" }] };
}
it("accepts natural independently attributed completed, remaining and failed reconnect reports", () => { const o = reconnect(); o.state.queryReply = "Website completed on the server while your browser was offline."; o.state.remainingPhase.reply = "The follow-up remains in progress and needs the human brief."; o.state.failedReconnectPhase.reply = "Riley could not finish: the provider rejected the configured model."; expect(gradeBrowserReconnect(o).passed).toBe(true); });
it("accepts actual server completion through a real browser disconnect", () => expect(gradeBrowserReconnect(reconnect()).passed).toBe(true));
it("rejects missing disconnect evidence", () => expect(gradeBrowserReconnect(undefined).passed).toBe(false));
it("rejects apparent success when the persisted worker failed", () => { const o = reconnect(); o.runs[0]!.status = "failed"; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects progress that happened before disconnection", () => { const o = reconnect(); o.state.workerStatusBeforeDisconnect = "succeeded"; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects a duplicate worker after reconnect", () => { const o = reconnect(); o.runs.push({ id: "duplicate", agentId: "morgan", status: "succeeded" }); expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects losing messages whose live stream events were missed", () => { const o = reconnect(); o.state.offlineCommentsRetained = false; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects a canonical answer only present in an earlier phase", () => { const o = reconnect(); o.reply = o.state.queryReply; o.state.queryReply = ""; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects a canonical answer split across more than one turn", () => { const o = reconnect(); o.state.canonicalRunIds = ["query", "extra"]; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects a remaining-work answer not bound to its own turn", () => { const o = reconnect(); o.state.remainingPhase.attributed = false; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects a remaining-work answer given after the work finished", () => { const o = reconnect(); o.state.remainingPhase.runStatusAtReply = "succeeded"; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects another disconnect without independent server activity", () => { const o = reconnect(); o.state.remainingPhase.runningWhileOffline = false; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects document title matches without actual latest-revision run provenance", () => { const o = reconnect(); o.revisions[0]!.createdByRunId = "other-run"; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects a failed-attempt question that no turn answered", () => { const o = reconnect(); o.state.failedReconnectPhase.reply = ""; o.state.failedReconnectPhase.attributed = false; expect(gradeBrowserReconnect(o).passed).toBe(false); });
it("rejects cancellation substituted for a genuine native failed phase", () => { const o = reconnect(); o.runs.find(run => run.id === "real-failure")!.status = "cancelled"; expect(gradeBrowserReconnect(o).passed).toBe(false); });