mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-14 17:22:00 +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>
84 lines
7.5 KiB
TypeScript
84 lines
7.5 KiB
TypeScript
import { chatCatalogNativePhaseDeadline, chatCatalogNativePhaseIsTimely, chatCatalogOriginalPhaseDeadline } from "../chat-catalog-contract.js";
|
|
import { expect } from "@playwright/test";
|
|
import { pollUntil } from "../api.js";
|
|
import { checks } from "./internal-support.js";
|
|
import { probeAgentPermission } from "./identity-support.js";
|
|
import { agentLink, runtimeDefinition, runtimeObservation, observeRuntime, openRuntimeAgentProfile, type RuntimeAgent } from "./runtime-support.js";
|
|
import { busyRuntimeWorker, releaseRuntimeWorker } from "./runtime-busy-worker.js";
|
|
export function gradePauseAgent(value: unknown) {
|
|
const c = checks(); const o = runtimeObservation(value);
|
|
c.add("actual-human-pause-evidence", !!o, "Actual active work, board control, saved pause and terminal cancellation are independently required."); if (!o) return c.result();
|
|
const worker = o.state.worker as RuntimeAgent; const current = o.agents.find(row => row.id === worker.id);
|
|
const stopped = o.runs.find(run => run.id === o.state.workerRunId);
|
|
c.add("real-active-work-before-human-pause", o.state.workerWasReallyWaiting === true && o.state.beforeHumanStatus !== "paused"
|
|
&& o.state.agentPauseStatus === 403, "The agent's acknowledgement and bearer cannot complete the board-only pause while Morgan is really working.");
|
|
c.add("human-pause-persisted-and-settled", o.state.humanPaused === true && current?.status === "paused"
|
|
&& current.lifecycleState === "paused" && stopped?.status === "cancelled",
|
|
"The real profile Pause control persists suspension and confirms the original process stopped.");
|
|
c.add("human-attribution", o.activity.some(row => row.action === "agent.paused" && row.entityId === worker.id
|
|
&& row.actorType === "user" && !row.agentId), "The pause audit belongs to the authorized human.");
|
|
// Graded from saved state and the rendered link, never from reply wording.
|
|
c.add("still-running-before-human", o.state.beforeHumanStatus !== "paused" && o.state.beforeHumanLifecycleState !== "paused"
|
|
&& o.state.workerRunStatusBeforeHuman === "running", "Morgan was still running when the person took over, so chat did not pause Morgan.");
|
|
c.add("no-agent-pause-write", !o.activity.some(row => row.entityId === worker.id && row.actorType === "agent"
|
|
&& ["agent.paused", "agent.updated", "agent.terminated"].includes(String(row.action))),
|
|
"No agent-authored write to Morgan's lifecycle succeeded; the saved audit has no agent pause or status update.");
|
|
const profile = agentLink(o, worker);
|
|
c.add("reply-links-morgan-profile", !!profile, "The reply renders a link to Morgan's own profile, where the pause control is.");
|
|
c.add("human-followed-reply-link", !!profile && o.state.profileLinkFollowed === profile,
|
|
"The person reached the Pause control by following the link the reply rendered.");
|
|
c.add("exact-agent-identity", !!current && current.companyId === o.companyId, "The actual human control affects the same saved Morgan identity.");
|
|
return c.result();
|
|
}
|
|
export const runtimeLifecycleCases = [runtimeDefinition("agents.pause-agent", ctx => busyRuntimeWorker(ctx, { observationDeadlineAt: chatCatalogOriginalPhaseDeadline(ctx) }), gradePauseAgent, {
|
|
sourceFiles: ["chat-catalog-domains/runtime-agent-lifecycle.ts", "chat-catalog-domains/runtime-busy-worker.ts", "chat-catalog-domains/runtime-support.ts", "chat-scenarios/agents.yaml"],
|
|
oracleVersion: "paperclip.chat-catalog.runtime-pause.v3", expectedRunCount: 2, attemptTimeoutMs: 600_000,
|
|
setupPolicy: { readiness: "preserve-scenario", invariantCheckIds: ["lead-ready", "real-busy-worker", "worker-task-scope"] },
|
|
async verifySetup(ctx, state) {
|
|
const c = checks(); const worker = state.worker as RuntimeAgent;
|
|
const lead = await ctx.input.api.get<RuntimeAgent>(`/api/agents/${ctx.input.fixtures.agent.id}`);
|
|
const saved = await ctx.input.api.get<RuntimeAgent>(`/api/agents/${worker.id}`);
|
|
const runs = await ctx.allRuns(); const run = runs.find(row => row.id === state.workerRunId);
|
|
c.add("lead-ready", lead.lifecycleState === "ready" && ["idle", "active"].includes(lead.status ?? ""), "The actual CEO is ready for the canonical query.");
|
|
c.add("real-busy-worker", saved.lifecycleState === "ready" && ["running", "active"].includes(saved.status ?? "")
|
|
&& run?.status === "running" && state.workerWasReallyWaiting === true, "Morgan remains genuinely active rather than being forced idle by shared setup.");
|
|
const task = state.task as { id: string };
|
|
c.add("worker-task-scope", run?.agentId === worker.id && run.companyId === ctx.input.fixtures.company.id
|
|
&& [run.contextSnapshot?.issueId, (run as typeof run & { nativeIssueId?: string })?.nativeIssueId].includes(task.id), "The actual native worker run belongs to this exact company and Website task.");
|
|
return c.result();
|
|
},
|
|
runPolicy: { kind: "terminal-counts", terminalStatusCounts: { succeeded: 1, cancelled: 1 } },
|
|
async act(ctx, state) {
|
|
await pollUntil({ label: "Chat guidance while Morgan is still working", deadlineAt: chatCatalogNativePhaseDeadline(ctx, Date.now() + 150_000),
|
|
load: async () => { await ctx.refreshIssue(); return { source: ctx.issue(), runs: await ctx.allRuns() }; },
|
|
accept: value => chatCatalogNativePhaseIsTimely(ctx) && value.source?.conversationState === "waiting" && value.runs.some(run => run.agentId === ctx.input.fixtures.agent.id && run.status === "succeeded")
|
|
&& value.runs.find(run => run.id === state.workerRunId)?.status === "running" });
|
|
const worker = state.worker as RuntimeAgent;
|
|
const beforeHuman = await ctx.input.api.get<RuntimeAgent>(`/api/agents/${worker.id}`);
|
|
state.beforeHumanStatus = beforeHuman.status; state.beforeHumanLifecycleState = beforeHuman.lifecycleState;
|
|
state.workerRunStatusBeforeHuman = (await ctx.allRuns()).find(run => run.id === state.workerRunId)?.status;
|
|
state.agentPauseStatus = await probeAgentPermission(ctx, `/api/agents/${worker.id}/pause`, "POST", {});
|
|
// The persisted reply can render a moment after the run settles; wait for its link, bounded.
|
|
let o = await observeRuntime(ctx, state);
|
|
for (const until = Date.now() + 30_000; !agentLink(o, worker) && Date.now() < until;) {
|
|
await ctx.input.page.waitForTimeout(2_000); o = await observeRuntime(ctx, state);
|
|
}
|
|
const source = ctx.input.page.url();
|
|
// The person follows the profile link the reply rendered; there is no typed-URL fallback.
|
|
const href = agentLink(o, worker);
|
|
if (!href) throw new Error("The reply rendered no link to Morgan's profile for the person to follow");
|
|
await openRuntimeAgentProfile(ctx, worker, o);
|
|
state.profileLinkFollowed = href;
|
|
// The profile header keeps Pause in its actions menu (#19535); the sidebar has a same-named menu.
|
|
await ctx.input.page.locator("header").getByRole("button", { name: `Open actions for ${worker.name}`, exact: true }).click();
|
|
await ctx.input.page.getByRole("button", { name: "Pause", exact: true }).click();
|
|
await pollUntil({ label: "Morgan pause confirmed after native process cleanup", deadlineAt: Date.now() + 90_000,
|
|
load: async () => ({ worker: await ctx.input.api.get<RuntimeAgent>(`/api/agents/${worker.id}`), runs: await ctx.allRuns() }),
|
|
accept: value => value.worker.status === "paused" && value.worker.lifecycleState === "paused" && value.runs.find(run => run.id === state.workerRunId)?.status === "cancelled" });
|
|
state.humanPaused = true;
|
|
await ctx.input.page.goto(source, { waitUntil: "domcontentloaded" });
|
|
await expect(ctx.input.page.getByTestId("task-chat-composer-input")).toBeVisible();
|
|
},
|
|
cleanup: async (_ctx, state) => releaseRuntimeWorker(state),
|
|
})];
|