Files
PaperClipAI/tests/runner-e2e/chat-catalog-domains/external-wrong-resolver.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

61 lines
9.5 KiB
TypeScript

import {randomUUID} from "node:crypto";
import {expect} from "@playwright/test";
import type {ChatCatalogContext} from "../chat-catalog-contract.js";
import type {ConnectionIntentInteraction} from "../../../packages/shared/src/types/issue.js";
import {pollUntil} from "../api.js";
import {createAuthenticatedPerson,enrollPrerequisitePerson} from "./identity-auth.js";
import {openStandingChat,waitForReply} from "./identity-support.js";
import {definition,externalObservation,observeExternal,checks,root,verifyRenderedReply} from "./external-support.js";
import type {Entity} from "./internal-support.js";
const people=new WeakMap<ChatCatalogContext,Awaited<ReturnType<typeof createAuthenticatedPerson>>>();
export function gradeWrongResolver(value:unknown){
const c=checks(),o=externalObservation(value);c.add("real-addressed-request-and-other-person",!!o,"A real native-created addressed card, two authenticated people and independently observed rejected requests are required.");if(!o)return c.result();
const before=o.state.original as ConnectionIntentInteraction|undefined,after=o.state.after as ConnectionIntentInteraction|undefined,related=o.state.related as Entity|undefined;
c.add("actual-Alex-card-and-visible-related-task",!!before&&before.kind==="connection_intent"&&before.payload.serviceSlug==="notion"&&before.status==="pending"&&before.addresseeUserId===o.state.alexId&&o.state.otherId!==o.state.alexId&&!!o.state.otherId&&o.state.relatedVisible===true&&o.state.otherSessionConfirmed===true&&!!related&&o.runs.some(run=>run.id===before.sourceRunId&&run.contextSnapshot?.issueId===related.id&&run.agentId===o.leadId&&run.status==="succeeded"&&run.runtimeMode==="native"),"The initial request was authored by a genuine admitted model run responsible to Alex; another signed-in company member can see the related task.");
const denied=o.state.denied as Record<string,number>|undefined;
c.add("all-addressed-resolution-surfaces-deny-other-person",!!denied&&["setupOptions","phase","complete","decline"].every(key=>denied[key]===403)&&o.state.waitingVisible===true&&o.state.connectControlAbsent===true,"The actual other human is refused setup details, phase mutation, completion and decline. Their visible card exposes waiting identity rather than a usable authorization control.");
c.add("pending-work-and-identity-preserved",!!before&&!!after&&after.id===before.id&&after.kind===before.kind&&after.sourceRunId===before.sourceRunId&&after.payload?.serviceSlug===before.payload.serviceSlug&&after.addresseeUserId===before.addresseeUserId&&after.status==="pending"&&!after.result&&!after.resolvedByUserId&&!after.resolvedByAgentId&&o.state.relatedStatusUnchanged===true&&o.state.alexSourceRunCount===1&&o.connections.length===0&&o.state.canonicalSourceDifferent===true,"Unauthorized responses neither release the pending task nor create a connection, readdress the card, or schedule another Alex-source run.");
// Reply wording is not graded: the refused resolution routes and the unchanged pending card prove the boundary held.
return c.result();
}
export const wrongResolverExternalCases=[definition("connection-setup.wrong-person-resolves",async ctx=>{
const browser=ctx.input.page.context().browser(),secrets=ctx.input.privateSecrets;if(!browser||!secrets)throw new Error("Wrong connection resolver needs two real authenticated people and private session ownership");
const session=await ctx.input.api.get<{user:{id:string}}>("/api/auth/get-session");if(!session.user?.id||session.user.id==="local-board")throw new Error("Actual signed-in other member is required");
const alex=await createAuthenticatedPerson({api:ctx.input.api,browser,secrets,name:"Alex"});people.set(ctx,alex);
try{
await enrollPrerequisitePerson(ctx, alex);
for(const scope of [{scopeType:"company",scopeId:ctx.input.fixtures.company.id},{scopeType:"agent",scopeId:ctx.input.fixtures.agent.id}])await ctx.input.api.post(`${root(ctx)}/budgets/policies`,{...scope,amount:500,hardStopEnabled:true,isActive:true,unpricedUsagePolicy:"allow"});
if((await ctx.input.api.get<{connections:unknown[]}>(`${root(ctx)}/tools/connections`)).connections.length)throw new Error("Wrong resolver requires no existing Notion account");
const response=await alex.request(`${root(ctx)}/issues`,"POST",{title:"Alex's launch notes",status:"todo",assigneeAgentId:ctx.input.fixtures.agent.id,description:"Read my Notion and summarize the current launch notes. I have not connected my Notion account yet."});
if(!response.ok){await response.body?.cancel();throw new Error("Alex could not create the actual related task through their authorized public identity");}const related=await response.json() as Entity;
const prepared=await pollUntil({label:"actual admitted Alex connection request",deadlineAt:Date.now()+150000,intervalMs:1000,load:async()=>({cards:await ctx.input.api.get<ConnectionIntentInteraction[]>(`/api/issues/${related.id}/interactions`),runs:await ctx.allRuns()}),accept:value=>value.runs.length===1&&value.runs[0]?.status==="succeeded"&&value.cards.some(card=>card.kind==="connection_intent"&&card.payload.serviceSlug==="notion"&&card.status==="pending"&&card.addresseeUserId===alex.id),reject:value=>value.runs.length>1||value.runs.some(run=>["failed","cancelled","timed_out"].includes(run.status))?"Alex's prerequisite did not produce one successful native request":undefined});
const cards=prepared.cards.filter(card=>card.kind==="connection_intent"&&card.payload.serviceSlug==="notion");if(cards.length!==1)throw new Error("Alex's prerequisite produced ambiguous requests");const original=cards[0]!;
const before=await ctx.input.api.get<Entity>(`/api/issues/${related.id}`),chat=await openStandingChat(ctx);
await ctx.input.api.patch(`/api/issues/${chat.id}`,{description:`Alex has a pending Notion connection request on related task ${before.identifier??before.id}. The current person can view that task as another company member.`});
return {original,related:before,alexId:alex.id,otherId:session.user.id};
}catch(error){await alex.dispose();people.delete(ctx);throw error;}
},gradeWrongResolver,{
oracleVersion:"paperclip.chat-catalog.external.v2",authentication:"human-session",expectedRunCount:2,minimumRunCount:2,expectedHumanMessageCount:1,attemptTimeoutMs:540000,
sourceFiles:["chat-catalog-domains/external-wrong-resolver.ts","chat-catalog-domains/identity-auth.ts","chat-catalog-domains/identity-private-browser.ts","chat-catalog-domains/identity-private-browser-worker.ts","chat-catalog-domains/identity-wizard.ts","chat-catalog-domains/identity-support.ts","chat-catalog-domains/external-support.ts","chat-catalog-domains/internal-support.ts","../../server/src/routes/connection-intents.ts","../../server/src/services/connection-intents.ts","chat-scenarios/connection-setup.yaml"],
async act(ctx,state){
await waitForReply({...ctx,allRuns:async()=>(await ctx.allRuns()).filter(run=>run.contextSnapshot?.issueId===ctx.issue()?.id)});
const original=state.original as ConnectionIntentInteraction,related=state.related as Entity,source=ctx.input.page.url();
const current=await ctx.input.api.get<{user:{id:string}}>("/api/auth/get-session");state.otherSessionConfirmed=current.user.id===state.otherId&&current.user.id!==state.alexId;
await ctx.input.page.goto(`/${ctx.input.fixtures.company.issuePrefix}/issues/${related.id}`,{waitUntil:"domcontentloaded"});await expect(ctx.input.page.getByText(related.title!,{exact:true}).first()).toBeVisible();state.relatedVisible=true;
await expect(ctx.input.page.getByTestId(/^connection-(?:intent|authorization)-waiting$/).filter({hasText:"Waiting for Alex"})).toBeVisible();state.waitingVisible=true;
state.connectControlAbsent=await ctx.input.page.getByRole("button",{name:/^Connect (?:Notion|account)$/}).count()===0;
const denied:Record<string,number>={};
for(const [name,suffix,method,data] of [["setupOptions","setup-options","GET",undefined],["phase","phase","POST",{phase:"authorizing"}],["complete","complete","POST",{connectionId:randomUUID()}],["decline","decline","POST",{}]] as const){
const response=await ctx.input.page.context().request.fetch(`${ctx.input.api.baseURL}/api/connection-intents/${original.id}/${suffix}`,{method,headers:{Origin:ctx.input.api.baseURL},...data?{data}:{}});denied[name]=response.status();await response.dispose();
}
state.denied=denied;await ctx.input.page.goto(source,{waitUntil:"domcontentloaded"});
},
async observe(ctx,state){
const related=state.related as Entity,original=state.original as ConnectionIntentInteraction,o=await observeExternal(ctx,state),cards=await ctx.input.api.get<ConnectionIntentInteraction[]>(`/api/issues/${related.id}/interactions`),after=cards.find(card=>card.id===original.id),task=await ctx.input.api.get<Entity>(`/api/issues/${related.id}`);
state.after=after;state.relatedStatusUnchanged=task.status===related.status&&task.assigneeAgentId===related.assigneeAgentId;state.alexSourceRunCount=o.runs.filter(run=>run.contextSnapshot?.issueId===related.id).length;state.canonicalSourceDifferent=ctx.issue()?.id!==related.id;
return {...o,state};
},
async verifyResult(ctx,_state,value){await verifyRenderedReply(ctx,value);const o=externalObservation(value)!;const source=ctx.input.page.url();try{await ctx.input.page.goto(`/${ctx.input.fixtures.company.issuePrefix}/issues/${(o.state.related as Entity).id}`,{waitUntil:"domcontentloaded"});await expect(ctx.input.page.getByTestId(/^connection-(?:intent|authorization)-waiting$/).filter({hasText:"Waiting for Alex"})).toBeVisible();}finally{await ctx.input.page.goto(source,{waitUntil:"domcontentloaded"});}},
async cleanup(ctx){try{await people.get(ctx)?.dispose();}finally{people.delete(ctx);}},
})];