mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-14 00:26:28 +01:00
## 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>
124 lines
10 KiB
TypeScript
124 lines
10 KiB
TypeScript
import { createHash } from "node:crypto";
|
|
import type { ChatCatalogContext, ChatCatalogQueryAttachment } from "./chat-catalog-contract.js";
|
|
import { expect } from "@playwright/test";
|
|
import { pollUntil } from "./api.js";
|
|
|
|
export interface ChatQueryAttachmentReceipt {
|
|
id: string; companyId: string; issueId: string; issueCommentId: string | null;
|
|
originalFilename: string | null; contentPath: string; byteSize: number; sha256: string;
|
|
createdByAgentId: string | null; createdByUserId: string | null; originatingRunId?: string | null;
|
|
}
|
|
export interface ChatQueryHumanComment {
|
|
id: string; body: string; authorAgentId?: string | null; authorUserId?: string | null;
|
|
issueId?: string; companyId?: string; createdByRunId?: string | null;
|
|
}
|
|
const factualFiles: Readonly<Record<string, string>> = {
|
|
"developer-labs.gated-lab": "historical-lab-link.md",
|
|
"documents-artifacts.resolve-feedback": "selected-source.md",
|
|
"documents-artifacts.restore-document": "selected-source.md",
|
|
"documents-artifacts.produce-download": "selected-source.md",
|
|
"documents-artifacts.revise-delivery": "selected-source.md",
|
|
"skills.author-skill": "skill-context.md", "skills.studio-test": "skill-context.md",
|
|
"skills.review-update": "skill-context.md", "skills.find-and-install": "skill-context.md", "skills.fork-read-only": "skill-context.md",
|
|
"connection-setup.secure-api-key": "provider-account.md",
|
|
"status-cards.explain-card": "selected-status-card.md", "status-cards.archive-card": "selected-status-card.md",
|
|
"routines.webhook-trigger": "selected-routine.md",
|
|
"cases.create-case": "customer-bug-brief.md", "app-permissions.ask-before-action": "launch-update.md",
|
|
};
|
|
const permitted = new Set(["steering.preserve-draft", "tasks.attachment-brief", "companies.import-preview", ...Object.keys(factualFiles)]);
|
|
export const factualContextQueryAttachment = (filename: string): readonly ChatCatalogQueryAttachment[] => [{ stateKey: "factualContextAttachment", filename }];
|
|
|
|
/** Upload factual user input through the real composer. This creates neither a
|
|
* comment nor a run; the unchanged canonical query submits the production file
|
|
* reference and attachmentId together. Contents are owned by the exact case. */
|
|
export async function uploadFactualQueryAttachment(ctx: ChatCatalogContext, input: { filename: string; body: string }): Promise<ChatQueryAttachmentReceipt> {
|
|
if (factualFiles[ctx.scenario.id] !== input.filename) throw new Error("Undeclared factual query attachment filename/scenario");
|
|
const bytes = Buffer.from(input.body, "utf8");
|
|
if (!bytes.length || bytes.length > 65_536) throw new Error("Factual context must be a bounded nonempty UTF-8 document");
|
|
const { api, page, fixtures } = ctx.input;
|
|
const health = await api.get<{ deploymentMode: string }>("/api/health");
|
|
const session = await api.get<{ user?: { id: string } }>("/api/auth/get-session");
|
|
if (health.deploymentMode !== "authenticated" || !session.user?.id || session.user.id === "local-board")
|
|
throw new Error("Factual query input requires its real authenticated human uploader");
|
|
await api.post(`/api/companies/${fixtures.company.id}/chats/${fixtures.agent.id}`);
|
|
await ctx.refreshIssue(); const source = ctx.issue();
|
|
if (!source) throw new Error("Factual upload requires the actual canonical standing conversation");
|
|
await page.goto(new URL(`/${fixtures.company.issuePrefix}/chats/${fixtures.agent.id}`, api.baseURL).href, { waitUntil: "domcontentloaded" });
|
|
await expect(page.getByTestId("task-chat-composer-input")).toBeVisible();
|
|
const endpoint = `/api/issues/${source.id}/attachments`;
|
|
if ((await api.get<ChatQueryAttachmentReceipt[]>(endpoint)).some(row => row.originalFilename === input.filename))
|
|
throw new Error("Factual input already exists; do not silently replace or duplicate source context");
|
|
await page.locator('input[type="file"]').first().setInputFiles({ name: input.filename, mimeType: "text/markdown", buffer: bytes });
|
|
const rows = await pollUntil({ label: "Actual human factual context upload", deadlineAt: Date.now() + 30_000,
|
|
load: () => api.get<ChatQueryAttachmentReceipt[]>(endpoint), accept: rows => rows.filter(row => row.originalFilename === input.filename).length === 1 });
|
|
const saved = rows.find(row => row.originalFilename === input.filename)!;
|
|
const expected = { ...saved, companyId: fixtures.company.id, issueId: source.id, createdByUserId: session.user.id, sha256: digest(bytes), byteSize: bytes.length };
|
|
const downloaded = await api.getAttachmentContent(saved.id);
|
|
if (!attachmentReceiptMatches({ expected, saved, bytes: downloaded, companyId: fixtures.company.id, issueId: source.id }) || !Buffer.from(downloaded).equals(bytes))
|
|
throw new Error("Factual upload bytes/source/human attribution do not match the original input");
|
|
const tray = page.getByTestId("task-chat-composer-attachments");
|
|
await expect(tray).toContainText(input.filename); await expect(tray).not.toContainText("Uploading");
|
|
const receipt = receiptProjection(saved);
|
|
await ctx.input.evidence("chat-catalog-factual-query-input.json", { scenarioId: ctx.scenario.id, receipt, body: input.body,
|
|
delivery: "pending canonical human comment attachment; description readback is not model delivery" });
|
|
return receipt;
|
|
}
|
|
function receiptProjection(row: ChatQueryAttachmentReceipt): ChatQueryAttachmentReceipt {
|
|
return { id: row.id, companyId: row.companyId, issueId: row.issueId, issueCommentId: row.issueCommentId,
|
|
originalFilename: row.originalFilename, contentPath: row.contentPath, byteSize: row.byteSize, sha256: row.sha256,
|
|
createdByAgentId: row.createdByAgentId, createdByUserId: row.createdByUserId, originatingRunId: row.originatingRunId };
|
|
}
|
|
const digest = (bytes: Uint8Array) => createHash("sha256").update(bytes).digest("hex");
|
|
/** Exact IssueChatThread submittedBody formatter; no alternate markdown or body rewriting. */
|
|
export function chatQueryWithAttachments(query: string, receipts: readonly ChatQueryAttachmentReceipt[]): string {
|
|
if (receipts.length > 1) throw new Error("Only one declared production attachment is supported");
|
|
return [query.trim(), ...receipts.map(item => `[${item.originalFilename!.replace(/[[\]]/g, "\\$&")}](${item.contentPath})`)].filter(Boolean).join("\n\n");
|
|
}
|
|
export function attachmentReceiptMatches(input: {
|
|
expected: ChatQueryAttachmentReceipt; saved: ChatQueryAttachmentReceipt; bytes: Uint8Array;
|
|
companyId: string; issueId: string; comment?: ChatQueryHumanComment;
|
|
}): boolean {
|
|
const { expected, saved, bytes, companyId, issueId, comment } = input;
|
|
return /^[a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12}$/i.test(saved.id)
|
|
&& saved.id === expected.id && saved.companyId === companyId && saved.issueId === issueId
|
|
&& saved.originalFilename === expected.originalFilename && !!saved.originalFilename
|
|
&& saved.contentPath === `/api/attachments/${saved.id}/content` && saved.contentPath === expected.contentPath
|
|
&& typeof saved.createdByUserId === "string" && !!saved.createdByUserId && saved.createdByUserId === expected.createdByUserId
|
|
&& saved.createdByAgentId == null && saved.originatingRunId == null
|
|
&& /^[a-f0-9]{64}$/.test(saved.sha256) && saved.sha256 === expected.sha256
|
|
&& digest(bytes) === saved.sha256 && saved.byteSize === bytes.byteLength && saved.byteSize === expected.byteSize
|
|
&& (comment ? saved.issueCommentId === comment.id && comment.authorUserId === saved.createdByUserId
|
|
&& comment.authorAgentId == null && comment.createdByRunId == null && comment.issueId === issueId && comment.companyId === companyId
|
|
: saved.issueCommentId == null);
|
|
}
|
|
export async function resolveChatQueryAttachments(ctx: ChatCatalogContext, specs: readonly ChatCatalogQueryAttachment[] = [], state: Record<string, unknown>) {
|
|
if (!specs.length) return [];
|
|
if (!permitted.has(ctx.scenario.id) || specs.length !== 1 || new Set(specs.map(row => row.stateKey)).size !== specs.length) throw new Error("Undeclared or duplicate catalog query attachment");
|
|
if (factualFiles[ctx.scenario.id] && specs[0]?.filename !== factualFiles[ctx.scenario.id]) throw new Error("Factual query attachment does not match its exact scenario filename");
|
|
if (ctx.scenario.id === "companies.import-preview" && (specs[0]?.filename !== "company-package.json" || specs[0]?.stateKey !== "packageAttachment"))
|
|
throw new Error("Company preview requires its exact declared package attachment");
|
|
await ctx.refreshIssue(); const issue = ctx.issue(); if (!issue) throw new Error("UI upload has no persisted source issue");
|
|
const saved = await ctx.input.api.get<ChatQueryAttachmentReceipt[]>(`/api/issues/${issue.id}/attachments`);
|
|
const receipts: ChatQueryAttachmentReceipt[] = [];
|
|
for (const spec of specs) {
|
|
const expected = state[spec.stateKey] as ChatQueryAttachmentReceipt | undefined;
|
|
const matches = saved.filter(row => row.id === expected?.id);
|
|
if (!expected || matches.length !== 1 || expected.originalFilename !== spec.filename || !spec.filename.trim() || /[\r\n\0]/.test(spec.filename)
|
|
|| receipts.some(row => row.id === expected.id)) throw new Error("Declared attachment differs from the persisted UI upload");
|
|
const receipt = matches[0]!; const bytes = await ctx.input.api.getAttachmentContent(receipt.id);
|
|
if (!attachmentReceiptMatches({ expected, saved: receipt, bytes, companyId: ctx.input.fixtures.company.id, issueId: issue.id })) throw new Error("Attachment source, content hash or human upload attribution does not match");
|
|
receipts.push(receiptProjection(receipt));
|
|
}
|
|
return receipts;
|
|
}
|
|
export async function verifyChatQueryAttachments(ctx: ChatCatalogContext, receipts: readonly ChatQueryAttachmentReceipt[], comment: ChatQueryHumanComment) {
|
|
const issue = ctx.issue(); if (!issue) return false;
|
|
const saved = await ctx.input.api.get<ChatQueryAttachmentReceipt[]>(`/api/issues/${issue.id}/attachments`);
|
|
for (const expected of receipts) {
|
|
const matches = saved.filter(row => row.id === expected.id); if (matches.length !== 1) return false;
|
|
const receipt = matches[0]!; const bytes = await ctx.input.api.getAttachmentContent(receipt.id);
|
|
if (!attachmentReceiptMatches({ expected, saved: receipt, bytes, companyId: ctx.input.fixtures.company.id, issueId: issue.id, comment })) return false;
|
|
}
|
|
return true;
|
|
}
|