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

106 lines
10 KiB
TypeScript

import { expect } from "@playwright/test";
import { pollUntil } from "../api.js";
import { readRegisteredArtifacts } from "../registered-artifact.js";
import { checks, sameTask, type Entity } from "./internal-support.js";
import { runtimeDefinition, runtimeWorker, runtimeObservation, observeRuntime } from "./runtime-support.js";
import type { ChatCatalogContext } from "../chat-catalog-contract.js";
interface WorkspaceView { workspace: { id: string; companyId: string; projectId?: string | null; name: string; cwd: string }; repositories: unknown[] }
export async function enableTaskWorkspaceView(ctx: ChatCatalogContext) {
await ctx.input.api.patch("/api/instance/settings/experimental", { enableIsolatedWorkspaces: true });
const settings = await ctx.input.api.get<{ enableIsolatedWorkspaces: boolean }>("/api/instance/settings/experimental");
if (settings.enableIsolatedWorkspaces !== true) throw new Error("The supported workspace view is not enabled");
}
export async function savedTaskWorkspace(ctx: ChatCatalogContext, options: { artifact?: boolean; workerName?: string; html?: boolean } = {}) {
const { execution, fixtures, workspacePath, nonce, api } = ctx.input;
const base = execution.profile.buildAgent({ environmentId: fixtures.environment.id, environmentFixtureId: execution.environment.id,
workspacePath, secretRefs: fixtures.secretRefs, executionId: nonce });
const { cwd: _cwd, ...taskConfig } = base.adapterConfig as Record<string, unknown>;
const worker = await runtimeWorker(ctx, options.workerName ?? "Morgan", { adapterConfig: taskConfig });
const company = `/api/companies/${fixtures.company.id}`;
for (const scope of [{ scopeType: "company", scopeId: fixtures.company.id }, { scopeType: "agent", scopeId: worker.id }])
await api.post(`${company}/budgets/policies`, { ...scope, amount: 500, hardStopEnabled: true, isActive: true, unpricedUsagePolicy: "allow" });
const marker = `Saved website prerequisite ${nonce}`;
const previousRuns = await ctx.allRuns();
const task = await api.post<Entity>(`${company}/issues`, { title: "Website task", status: "todo", assigneeAgentId: worker.id,
description: `Create a local file website-fixture.txt containing exactly ${JSON.stringify(marker)} in your admitted task workspace. ${options.html ? `Also save index.html as a semantic HTML page with that same visible reference in an h1 and a short Acorn product paragraph. Do not start any preview server yet.` : ""} Save a brief task document as evidence, ${options.artifact ? "register the actual website-fixture.txt file as a downloadable artifact work product titled website-fixture.txt," : ""} then complete this fixture task normally. Do not create a project, repository, another task, or remote side effect.` });
await pollUntil({ label: "Real projectless website fixture completion", deadlineAt: Date.now() + 180_000,
load: async () => ({ task: await api.get<Entity>(`/api/issues/${task.id}`), runs: (await ctx.allRuns()).filter(run => !previousRuns.some(prior => prior.id === run.id)) }),
accept: value => value.task.status === "done" && value.runs.length === 1 && value.runs.every(run => run.status === "succeeded"),
reject: value => value.runs.some(run => ["failed", "cancelled", "timed_out"].includes(run.status)) ? "Projectless prerequisite provider run failed" : undefined });
const view = await api.get<WorkspaceView>(`/api/issues/${task.id}/workspace`);
const file = await api.get<{ resource?: { workspaceKind?: string }; content?: { data?: string } }>(`/api/issues/${task.id}/file-resources/content?workspace=execution&path=website-fixture.txt`);
if (view.workspace.projectId || file.resource?.workspaceKind !== "execution_workspace" || file.content?.data?.trim() !== marker)
throw new Error("Projectless fixture has no actual task-bound saved file");
const fixtureRunId = (await ctx.allRuns()).find(run => !previousRuns.some(prior => prior.id === run.id))!.id;
const [artifact] = options.artifact ? await readRegisteredArtifacts(api, task.id, fixtureRunId, ["website-fixture.txt"]) : [];
return { worker, task: await api.get<Entity>(`/api/issues/${task.id}`), workspaceBefore: view, marker, artifact,
fileBefore: file.content.data, fixtureRunId };
}
export function gradeLocateFiles(value: unknown) {
const c = checks(); const o = runtimeObservation(value);
c.add("actual-task-workspace-evidence", !!o, "The actual admitted binding, file bytes and setup run identity are independently required."); if (!o) return c.result();
const before = o.state.workspaceBefore as WorkspaceView | undefined; const after = o.state.workspaceAfter as WorkspaceView | undefined;
const task = o.state.task as Entity | undefined;
c.add("same-projectless-task-binding", !!before?.workspace?.id && !!after && before.workspace.id === after.workspace.id
&& before.workspace.cwd === after.workspace.cwd && after.workspace.companyId === o.companyId && after.workspace.projectId == null
&& !!task && sameTask(task, o.tasks.find(row => row.id === task.id)), "The correct projectless binding persists without reassigning or changing the task.");
c.add("actual-saved-files-retained", o.state.fileAfter === o.state.fileBefore && typeof o.state.fileBefore === "string"
&& o.state.fileBefore.trim() === o.state.marker && o.runs.some(run => run.id === o.state.fixtureRunId && run.agentId === (o.state.worker as Entity).id
&& run.status === "succeeded"), "A genuine worker created the prerequisite file and it remains unchanged.");
// The person needs to reach the bound workspace, so the reply must render a link to its page; the
// scripted human then clicks that link. A cwd or id copied as text is not graded: the request did not ask
// for a literal, and abbreviated paths are a valid way to name a location the link already identifies.
const prefix = String(o.state.companyPrefix);
c.add("bound-workspace-page-link", !!after && o.links.some(href => workspaceLink(o.origin, href, after.workspace.id, prefix)),
"The rendered reply links the supported view of the workspace actually bound to the task.");
c.add("no-other-workspace-link", !!after && workspaceResultLinks(o.reply, o.links)
.filter(href => { try { return new URL(href, o.origin).pathname.includes("/execution-workspaces/"); } catch { return false; } })
.every(href => workspaceLink(o.origin, href, after.workspace.id, prefix)),
"Every workspace link in the reply identifies the bound workspace, not an agent default or another task's workspace.");
return c.result();
}
export function workspaceResultLinks(reply: string, rendered: string[]) {
return [...new Set([...rendered, ...Array.from(reply.matchAll(/\[[^\]]*\]\(([^)\s]+)\)/g), match => match[1]!)])];
}
export function workspaceLink(origin: string, href: string, id: string, prefix: string) {
try { const url = new URL(href, origin); const parts = url.pathname.split("/").map(decodeURIComponent); const at = parts.indexOf("execution-workspaces");
return url.origin === new URL(origin).origin && at === 2 && parts[1] === prefix && parts[at + 1] === id
&& (parts.length === 4 || parts.length === 5 && ["services", "configuration", "runtime-logs", "issues", "routines"].includes(parts[4]!)); } catch { return false; }
}
export const runtimeWorkspacesCases = [runtimeDefinition("workspaces.locate-files", async ctx => {
// The actual detail route is gated, independently of workspace realization.
// Enable its supported UI fixture before the query, never after a bad link.
if (!ctx.input.fixtures.company.issuePrefix) throw new Error("Workspace fixture has no public company prefix");
await enableTaskWorkspaceView(ctx);
return { ...await savedTaskWorkspace(ctx), companyPrefix: ctx.input.fixtures.company.issuePrefix };
}, gradeLocateFiles, {
sourceFiles: ["chat-catalog-domains/runtime-workspaces.ts", "chat-catalog-domains/runtime-support.ts", "chat-scenarios/workspaces.yaml", "../../ui/src/components/IsolatedWorkspacesRouteGate.tsx", "../../ui/src/pages/ExecutionWorkspaceDetail.tsx"],
expectedRunCount: 2, attemptTimeoutMs: 540_000, oracleVersion: "paperclip.chat-catalog.runtime.locate-files.v3",
async observe(ctx, state) {
const id = (state.task as Entity).id;
state.workspaceAfter = await ctx.input.api.get<WorkspaceView>(`/api/issues/${id}/workspace`);
const file = await ctx.input.api.get<{ content: { data: string } }>(`/api/issues/${id}/file-resources/content?workspace=execution&path=website-fixture.txt`);
state.fileAfter = file.content.data;
return observeRuntime(ctx, state);
},
async verifyResult(ctx, state, value) {
const o = runtimeObservation(value); if (!o) throw new Error("Workspace result missing");
const view = state.workspaceAfter as WorkspaceView;
const prefix = state.companyPrefix;
if (typeof prefix !== "string" || !prefix) throw new Error("Workspace view lost its actual company prefix");
const links = workspaceResultLinks(o.reply, o.links);
if (links.some(link => { try { return new URL(link, o.origin).pathname.includes("/execution-workspaces/")
&& !workspaceLink(o.origin, link, view.workspace.id, prefix); } catch { return false; } }))
throw new Error("The actual reply supplied an incorrect workspace link");
// Follow only the link the reply rendered; a typed URL would pass a reply the person cannot act on.
const href = o.links.find(link => workspaceLink(o.origin, link, view.workspace.id, prefix));
if (!href) throw new Error("The reply rendered no link to the bound workspace page");
const source = ctx.input.page.url();
await ctx.input.page.locator(`[id^="comment-"] a[href=${JSON.stringify(href)}]`).first().click();
await expect(ctx.input.page).toHaveURL(new RegExp(`/${prefix}/execution-workspaces/${view.workspace.id}(?:[/?#]|$)`));
await expect(ctx.input.page.getByRole("heading", { name: view.workspace.name, exact: true }).first()).toBeVisible();
await ctx.input.evidence("runtime-task-workspace-opened.json", { taskId: (state.task as Entity).id, workspaceId: view.workspace.id, url: ctx.input.page.url() });
await ctx.input.page.goto(source, { waitUntil: "domcontentloaded" });
},
})];