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>
110 lines
8.5 KiB
TypeScript
110 lines
8.5 KiB
TypeScript
import { createHash } from "node:crypto";
|
|
import { expect } from "@playwright/test";
|
|
import { parseIssueReferenceFromHref } from "../../../ui/src/lib/issue-reference.js";
|
|
import type { ChatCatalogCase, ChatCatalogContext } from "../chat-catalog-contract.js";
|
|
import { pollUntil } from "../api.js";
|
|
import { internalResultLink, entity, hasLink, observeInternal, type Entity, type InternalObservation } from "./internal-support.js";
|
|
|
|
const digest = (body: string) => createHash("sha256").update(body).digest("hex");
|
|
export function savedDesignInstruction(o: InternalObservation, body: string, design: Entity) {
|
|
const valid: string[] = [];
|
|
const reference = (href: string) => {
|
|
let local = href;
|
|
try {
|
|
if (/^https?:/i.test(href)) {
|
|
const url = new URL(href); if (url.origin !== new URL(o.origin).origin) return "{{OTHERREF}}";
|
|
local = `${url.pathname}${url.search}${url.hash}`;
|
|
}
|
|
const segments = local.split(/[?#]/, 1)[0]!.split("/").filter(Boolean), issueAt = segments.indexOf("issues");
|
|
if (design.identifier && issueAt > 0 && segments[issueAt - 1] !== design.identifier.split("-")[0]) return "{{OTHERREF}}";
|
|
} catch { return "{{OTHERREF}}"; }
|
|
const parsed = parseIssueReferenceFromHref(local, design.identifier ? new Set([design.identifier.split("-")[0]!]) : undefined);
|
|
if (!parsed || !hasLink({ ...o, links: [parsed.href] }, "issues", design)) return "{{OTHERREF}}";
|
|
const token = `{{DESIGNREF${valid.length}}}`; valid.push(token); return token;
|
|
};
|
|
const visibleSource = body.replace(/^ {0,3}(`{3,}|~{3,})[^\n]*\n[\s\S]*?^ {0,3}\1[^\n]*$/gm, "")
|
|
.replace(/<!--[\s\S]*?-->/g, "").replace(/<[^>]*>[\s\S]*?<\/[^>]*>/g, "");
|
|
// Rendered-comment proof below is mandatory; lexical recognition alone never
|
|
// proves that a Markdown or bare identifier became a visible, usable link.
|
|
const text = visibleSource.replace(/\[[^\]]+\]\(<?([^\s)>]+)>?(?:\s+["'][^"']*["'])?\)/g, (_link, href: string) => reference(href))
|
|
.replace(/\b[A-Z][A-Z0-9]*-\d+\b/gi, identifier => reference(identifier));
|
|
if (!valid.length) return false;
|
|
const directive = /\b(?:follow|use|adhere\s+to|implement\s+(?:according\s+to|in\s+line\s+with)|build\s+(?:according\s+to|in\s+line\s+with))\b/i;
|
|
const negative = (prefix: string) => {
|
|
const t = prefix.replace(/\b(?:do not|don't) forget to\b/gi, "remember to");
|
|
return /\b(?:do not|don't|must not|should not|never|not to|not)\b[^{}.;!\n]{0,60}\b(?:follow|use|adhere|implement|build)\b[^{}.;!\n]{0,80}$/i.test(t)
|
|
|| /\b(?:avoid|ignore|disregard|stop)\b[^{}.;!\n]{0,80}$/i.test(t);
|
|
};
|
|
const clauses = text.split(/[.;!\n]+/).filter(clause => valid.some(token => clause.includes(token)));
|
|
if (clauses.some(clause => valid.some(token => { const at = clause.indexOf(token); return at >= 0 && negative(clause.slice(0, at)); }))) return false;
|
|
if (/\b(?:(?:do not|don't|never|must not|should not)\s+(?:follow|use|adhere\s+to)|ignore|disregard)\s+(?:it|this|that|the\s+(?:approved\s+)?design)\b/i.test(text)) return false;
|
|
return clauses.some(clause => directive.test(clause) && valid.some(token => {
|
|
const instruction = directive.exec(clause); return !!instruction && clause.indexOf(token) > instruction.index;
|
|
}));
|
|
}
|
|
interface ReferenceProof { commentId: string; taskId: string; designId: string; bodySha256: string; href: string; visible: boolean; destinationMatched: boolean; publicReadMatched: boolean }
|
|
export function validRenderedCommentReference(o: InternalObservation, comment: Entity, design: Entity) {
|
|
const proof = o.state.savedCommentReference as ReferenceProof | undefined;
|
|
return !!proof && proof.commentId === comment.id && comment.issueId === proof.taskId && proof.taskId === entity(o.state, "task").id && proof.designId === design.id
|
|
&& proof.bodySha256 === digest(comment.body ?? "") && proof.visible === true && proof.destinationMatched === true && proof.publicReadMatched === true
|
|
&& hasLink({ ...o, links: [proof.href] }, "issues", design);
|
|
}
|
|
export const RENDERED_REPLY_WAIT_MS = 60_000;
|
|
/** Waits for every saved agent reply of the conversation to render; throws a named error otherwise. */
|
|
export async function waitForRenderedReply(ctx: Pick<ChatCatalogContext, "issue" | "input">, o: Pick<InternalObservation, "leadId" | "runs">,
|
|
timeoutMs = RENDERED_REPLY_WAIT_MS, intervalMs = 500) {
|
|
const source = ctx.issue();
|
|
if (!source) throw new Error("Rendered reply wait: missing source conversation");
|
|
const chat = await ctx.input.api.get<Entity[]>(`/api/issues/${source.id}/comments?order=asc`);
|
|
const replies = chat.filter(row => row.authorAgentId === o.leadId && o.runs.some(run => run.id === row.createdByRunId));
|
|
if (!replies.length) throw new Error("Rendered reply wait: the settled source run saved no reply");
|
|
const deadlineAt = Date.now() + timeoutMs;
|
|
for (const reply of replies) {
|
|
const rendered = ctx.input.page.locator(`[id=${JSON.stringify(`comment-${reply.id}`)}]`).filter({ visible: true });
|
|
try {
|
|
await pollUntil({ label: `rendered reply ${reply.id}`, deadlineAt, intervalMs, load: async () => await rendered.count() > 0, accept: Boolean });
|
|
} catch (error) {
|
|
throw new Error(`Rendered reply wait: saved reply ${reply.id} did not render in the open conversation within ${timeoutMs}ms`, { cause: error });
|
|
}
|
|
}
|
|
}
|
|
export function withRenderedCommentReference(base: ChatCatalogCase): ChatCatalogCase {
|
|
return { ...base, oracleVersion: "paperclip.chat-catalog.internal.comment-with-reference.v2",
|
|
sourceFiles: [...base.sourceFiles, "chat-catalog-domains/internal-comment-reference.ts", "../../ui/src/lib/issue-reference.ts", "../../ui/src/components/MarkdownBody.tsx"],
|
|
async act(ctx, state) {
|
|
await pollUntil({ label: "Saved comment source turn", deadlineAt: Date.now() + 150000,
|
|
load: async () => { await ctx.refreshIssue(); return ctx.allRuns(); },
|
|
accept: runs => runs.length === 1 && runs[0]!.status === "succeeded" && ctx.issue()?.conversationState === "waiting",
|
|
reject: runs => runs.some(run => ["failed", "cancelled", "timed_out"].includes(run.status)) ? "Source run failed" : undefined });
|
|
const saved = await observeInternal(ctx, state), target = entity(state, "task"), design = entity(state, "design");
|
|
const added = saved.comments.filter(row => row.authorAgentId === saved.leadId && row.createdByRunId && saved.runs.some(run => run.id === row.createdByRunId));
|
|
if (added.length !== 1 || !savedDesignInstruction(saved, added[0]!.body ?? "", design)) return;
|
|
// Run settle precedes the reply's render in the open conversation; read
|
|
// rendered links only after the saved reply itself is on the page.
|
|
await waitForRenderedReply(ctx, saved);
|
|
const o = await observeInternal(ctx, state);
|
|
const taskHref = o.links.find(href => hasLink({ ...o, links: [href] }, "issues", target));
|
|
if (!taskHref) return;
|
|
const page = ctx.input.page, source = page.url();
|
|
try {
|
|
const link = internalResultLink(ctx, o, taskHref).first();
|
|
await expect(link).toBeVisible(); await link.click();
|
|
await page.waitForURL(url => hasLink({ ...o, links: [url.href] }, "issues", target));
|
|
await expect(page.getByRole("heading", { name: target.title!, exact: true }).first()).toBeVisible();
|
|
const comment = page.locator(`[id=${JSON.stringify(`comment-${added[0]!.id}`)}]`);
|
|
await expect(comment).toBeVisible();
|
|
const hrefs = await comment.locator("a[href]").evaluateAll(elements => elements.map(element => element.getAttribute("href")!));
|
|
const href = hrefs.find(value => hasLink({ ...o, links: [value] }, "issues", design));
|
|
if (!href) return;
|
|
const designLink = comment.locator(`a[href=${JSON.stringify(href)}]`).first(); await expect(designLink).toBeVisible(); await designLink.click();
|
|
await page.waitForURL(url => hasLink({ ...o, links: [url.href] }, "issues", design));
|
|
await expect(page.getByRole("heading", { name: design.title!, exact: true }).first()).toBeVisible();
|
|
const actual = await ctx.input.api.get<Entity>(`/api/issues/${design.id}`);
|
|
state.savedCommentReference = { commentId: added[0]!.id, taskId: target.id, designId: design.id, bodySha256: digest(added[0]!.body ?? ""), href,
|
|
visible: true, destinationMatched: true, publicReadMatched: actual.id === design.id && actual.companyId === o.companyId } satisfies ReferenceProof;
|
|
await ctx.input.evidence("chat-saved-comment-reference.json", state.savedCommentReference);
|
|
} finally { await page.goto(source, { waitUntil: "domcontentloaded" }); await expect(page.getByTestId("task-chat-composer-input")).toBeVisible(); }
|
|
},
|
|
};
|
|
}
|