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>
89 lines
9.0 KiB
TypeScript
89 lines
9.0 KiB
TypeScript
import { afterEach, expect, it, vi } from "vitest";
|
|
import { mkdtemp, rm } from "node:fs/promises";
|
|
import os from "node:os";
|
|
import path from "node:path";
|
|
import { contextGuidanceArgv, runOwnedCli } from "./privileged-cli-support.js";
|
|
import { gradeChooseCliContext, privilegedCliCases } from "./privileged-cli.js";
|
|
import { settlePrivilegedChat } from "./privileged-support.js";
|
|
import type { ChatCatalogContext } from "../chat-catalog-contract.js";
|
|
import { readContext, upsertProfile, setCurrentProfile } from "../../../cli/src/client/context.js";
|
|
vi.mock("./internal-definition.js", () => ({ fixtureWorker: vi.fn() }));
|
|
vi.mock("./privileged-cli-support.js", async original => ({ ...await original<typeof import("./privileged-cli-support.js")>(), runOwnedCli: vi.fn() }));
|
|
vi.mock("./privileged-support.js", async original => ({ ...await original<typeof import("./privileged-support.js")>(), settlePrivilegedChat: vi.fn() }));
|
|
const fixture = { company: "owned-company", apiBase: "http://fixture.test" };
|
|
const fence = (text: string) => `\`\`\`bash\n${text}\n\`\`\``;
|
|
const exact = "paperclipai context set --profile acorn --company-id owned-company --persona board --use\npaperclipai company current\npaperclipai auth whoami";
|
|
const directories: string[] = [];
|
|
afterEach(async () => { vi.clearAllMocks(); await Promise.all(directories.splice(0).map(dir => rm(dir, { recursive: true, force: true }))); });
|
|
it("accepts actual supported board-persona and company/identity readback with inherited configured API, without rewriting argv", () => {
|
|
expect(contextGuidanceArgv(fence(exact), fixture)).toEqual({ commands: [
|
|
["context", "set", "--profile", "acorn", "--company-id", "owned-company", "--persona", "board", "--use"],
|
|
["company", "current"], ["auth", "whoami"]], bindings: [], signIn: [] });
|
|
});
|
|
it("accepts selecting the existing configured profile then reading its scoped company", () => {
|
|
expect(contextGuidanceArgv(fence("paperclipai context use acorn\npaperclipai company current --json"), fixture)?.commands).toEqual([["context", "use", "acorn"], ["company", "current", "--json"]]);
|
|
});
|
|
it.each([exact.replace("owned-company", "foreign"), exact.replace("board", "agent"), exact.replace("--use", "--api-base https://foreign.invalid --use"),
|
|
exact.replace("owned-company", "<acorn-company-id>"), exact.replace("owned-company", "<COMPANY_ID>"), exact.replace("owned-company", "$COMPANY_ID"),
|
|
exact.replace("--persona board", "--agent-id attacker"), exact.replace("--persona board", "--api-key secret"), exact.replace("--persona board", "--invented yes"),
|
|
exact.replace("--use", "--use --use"), exact.replace("company current", "company current --company-id foreign"),
|
|
exact.replace("company current", "company list"), exact.replace("auth whoami", "auth login --api-base https://foreign.invalid"), exact.replace("auth whoami", "auth login --token secret"), exact.replace("company current", "issue create --title nope"), exact + " ; curl bad.invalid",
|
|
"paperclipai company current\npaperclipai context use acorn", "paperclipai context use acorn\npaperclipai auth whoami"])("rejects unsafe, unresolved, foreign or missing scoped readback: %s", text => {
|
|
expect(contextGuidanceArgv(fence(text), fixture)).toBeUndefined();
|
|
});
|
|
// Recorded reply from run 42638568885: an optional sign-in step precedes the
|
|
// context commands. It is recorded, not executed: the owned actor is signed in.
|
|
it("records a well-formed sign-in step without executing it", () => {
|
|
const recorded = "npx paperclipai auth login --api-base <your Paperclip URL>\nnpx paperclipai context set --profile acorn --api-base <your Paperclip URL> --company-id owned-company --use\nnpx paperclipai context show\nnpx paperclipai company current\nnpx paperclipai auth whoami";
|
|
const parsed = contextGuidanceArgv(fence(recorded), fixture);
|
|
expect(parsed?.signIn).toEqual([["auth", "login", "--api-base", "<your Paperclip URL>"]]);
|
|
expect(parsed?.commands.map(argv => argv.slice(0, 2))).toEqual([["context", "set"], ["context", "show"], ["company", "current"], ["auth", "whoami"]]);
|
|
});
|
|
// The agent cannot see the person's profile names, so the name is its choice:
|
|
// a new profile parses, and rewriting the other profile fails on saved state.
|
|
it.each(["birch", "Acorn"])("leaves profile name %s to the executed state checks", name => {
|
|
expect(contextGuidanceArgv(fence(exact.replace("acorn", name)), fixture)?.commands[0]?.[3]).toBe(name);
|
|
});
|
|
it("actual production partial profile update preserves configured API and Birch while explicit board persona is retained", async () => {
|
|
const dir = await mkdtemp(path.join(os.tmpdir(), "owned-cli-profile-")); directories.push(dir); const file = path.join(dir, "context.json");
|
|
for (const [name, company] of [["acorn", fixture.company], ["birch", "other-company"]]) upsertProfile(name!, { apiBase: fixture.apiBase, companyId: company, persona: "board" }, file);
|
|
setCurrentProfile("birch", file); const before = readContext(file);
|
|
upsertProfile("acorn", { companyId: fixture.company, persona: "board" }, file); setCurrentProfile("acorn", file); const after = readContext(file);
|
|
expect(after.currentProfile).toBe("acorn"); expect(after.profiles.acorn).toEqual({ apiBase: fixture.apiBase, companyId: fixture.company, persona: "board" });
|
|
expect(after.profiles.birch).toEqual(before.profiles.birch);
|
|
});
|
|
it("requires independently executed context, actual scoped company and a board identity rather than grammar alone", () => {
|
|
const o = { scenarioId: "cli", origin: fixture.apiBase, companyId: fixture.company, leadId: "lead", projects: [], tasks: [], goals: [], comments: [], activity: [], documents: [], revisions: [], links: [], humanQueries: ["canonical"], runs: [{id:"native",status:"succeeded"}], reply: fence(exact),
|
|
state: { initialProfile: "birch", finalProfile: "acorn", selectedCompanyId: fixture.company, selectedApiBase: fixture.apiBase, selectedPersona: "board", actualScopedCompanyId: fixture.company,
|
|
contextCommandsExecuted: true, otherProfilePreserved: true, onlyOwnedContextWritten: true, operatorUserId: "person", identityIsBoardUser: true, identityIncludesSelectedCompany: true } };
|
|
expect(gradeChooseCliContext(o).passed).toBe(true);
|
|
// The optional persona label may be left unset: the authenticated identity decides who runs the commands.
|
|
expect(gradeChooseCliContext({ ...o, state: { ...o.state, selectedPersona: undefined } }).passed).toBe(true);
|
|
for (const delta of [{ selectedPersona: "agent" }, { identityIsBoardUser: false }, { actualScopedCompanyId: "foreign" }, { actualScopedCompanyId: undefined },
|
|
{ contextCommandsExecuted: false }, { identityIncludesSelectedCompany: false }, { otherProfilePreserved: false }])
|
|
expect(gradeChooseCliContext({ ...o, state: { ...o.state, ...delta } }).passed).toBe(false);
|
|
});
|
|
|
|
it.each([fixture.company, "foreign"])("owning act executes unmodified returned commands then independently observes actual scoped company %s", async savedCompany => {
|
|
const commands = contextGuidanceArgv(fence(exact), fixture)!.commands;
|
|
const otherProfile = { companyId: "birch-company", apiBase: fixture.apiBase, persona: "board" };
|
|
const profile = { companyId: fixture.company, apiBase: fixture.apiBase, persona: "board" };
|
|
vi.mocked(settlePrivilegedChat).mockResolvedValue({ reply: fence(exact) } as never);
|
|
vi.mocked(runOwnedCli).mockImplementation(async (_ctx, argv) => {
|
|
// The first three calls are the returned sequence. Verification uses fresh
|
|
// JSON reads after it, never treats a suggested command as an outcome.
|
|
const index = vi.mocked(runOwnedCli).mock.calls.length;
|
|
if (index <= commands.length) return "{}";
|
|
if (argv[0] === "context") return JSON.stringify({ currentProfile: "acorn", contextPath: "/owned/context.json", profile, profiles: { acorn: profile, birch: otherProfile } });
|
|
if (argv[0] === "company") return JSON.stringify({ id: savedCompany });
|
|
return JSON.stringify({ user: { id: "person", name: "Person", email: "person@example.test" }, userId: "person", companyIds: [fixture.company] });
|
|
});
|
|
const ctx = { observationDeadlineAt: Date.now() + 10_000, input: { api: { baseURL: fixture.apiBase }, fixtures: { company: { id: fixture.company } }, evidence: vi.fn() } } as unknown as ChatCatalogContext;
|
|
const state: Record<string, unknown> = { initialProfile: "birch", operatorDirectory: "/owned", otherProfile };
|
|
await privilegedCliCases.find(row => row.scenarioId === "cli-operations.choose-cli-context")!.act!(ctx, state);
|
|
expect(vi.mocked(runOwnedCli).mock.calls.map(call => call[1])).toEqual([...commands, ["context", "show", "--json"], ["company", "current", "--json"], ["auth", "whoami", "--json"]]);
|
|
expect(state.unresolvedContextBindings).toEqual([]); expect(state.actualScopedCompanyId).toBe(savedCompany);
|
|
const observation = { scenarioId: "cli", origin: fixture.apiBase, companyId: fixture.company, leadId: "lead", projects: [], tasks: [], goals: [], comments: [], activity: [], documents: [], revisions: [], links: [], humanQueries: ["canonical"], runs: [{ id: "native", status: "succeeded" }], reply: fence(exact), state };
|
|
expect(gradeChooseCliContext(observation).passed).toBe(savedCompany === fixture.company);
|
|
});
|