Files
PaperClipAI/tests/runner-e2e/chat-vault-private-server.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

255 lines
8.3 KiB
TypeScript

import { reviewedChatExecutionIds } from "./chat-runtime-profile.js";
import { constants } from "node:fs";
import { lstat, realpath, open, writeFile } from "node:fs/promises";
import path from "node:path";
import { CHAT_PRIVATE_INPUTS_ENV } from "./chat-private-inputs.js";
export const vaultScenarioId = "secrets.vault-reference";
export const vaultExecutionId = reviewedChatExecutionIds(vaultScenarioId)[0]!;
export const vaultExecutionIds = reviewedChatExecutionIds(vaultScenarioId);
export const vaultPrivateFields = [
"provider",
"region",
"deploymentId",
"kmsKeyId",
"secretReference",
"accessKeyIdEnv",
"secretAccessKeyEnv",
"sessionTokenEnv",
"expectedValueEnv",
] as const;
type Binding = {
disposable: boolean;
resourceIds: string[];
credentialEnv: string[];
prerequisites: Record<string, string>;
};
interface Envelope {
version: number;
executionId: string;
references: Record<string, string>;
external?: { schemaVersion?: number; providers?: { vault?: Binding } };
declaredReferenceNames?: string[];
declaredExternalRequirements?: Array<{ provider: string; fields: string[] }>;
setupBlockers?: string[];
}
export function privateVaultMetadataMatches(
info: { uid: number; mode: number },
expectedUid: number,
) {
return info.uid === expectedUid && !(info.mode & 0o077);
}
export function validVaultBinding(
binding: Binding | undefined,
): binding is Binding {
if (
!binding ||
binding.disposable !== true ||
!Array.isArray(binding.resourceIds) ||
!Array.isArray(binding.credentialEnv) ||
!binding.prerequisites
)
return false;
const p = binding.prerequisites;
if (
vaultPrivateFields.some(
(field) => typeof p[field] !== "string" || !p[field].trim(),
) ||
p.provider !== "aws_secrets_manager" ||
!/^[a-z]{2}(?:-[a-z]+)+-\d$/.test(p.region!) ||
!/^[a-zA-Z0-9][a-zA-Z0-9_.-]{0,80}$/.test(p.deploymentId!)
)
return false;
const secret =
/^arn:(aws(?:-us-gov|-cn)?):secretsmanager:([^:]+):(\d{12}):secret:([^\r\n]+)$/.exec(
p.secretReference!,
);
const kms =
/^arn:(aws(?:-us-gov|-cn)?):kms:([^:]+):(\d{12}):key\/([a-f0-9-]{36})$/.exec(
p.kmsKeyId!,
);
if (
!secret ||
!kms ||
secret[1] !== kms[1] ||
secret[2] !== p.region ||
kms[2] !== p.region ||
secret[3] !== kms[3] ||
![p.secretReference, p.kmsKeyId].every((ref) =>
binding.resourceIds.includes(ref!),
)
)
return false;
const names = [
p.accessKeyIdEnv!,
p.secretAccessKeyEnv!,
p.sessionTokenEnv!,
p.expectedValueEnv!,
];
return (
new Set(names).size === names.length &&
names.every(
(name) =>
/^PAPERCLIP_CHAT_EVAL_[A-Z0-9_]+$/.test(name) &&
binding.credentialEnv.includes(name),
) &&
binding.credentialEnv.every((name) => names.includes(name))
);
}
/** Only the selected product server receives these credentials. Oracle values,
* config pointers, profiles and SDK fallback sources never cross this boundary. */
export async function readSelectedVaultServerEnvironment(input: {
executionIds: readonly string[];
temporaryRoot: string;
backendOrigin: string;
env: NodeJS.ProcessEnv;
}): Promise<NodeJS.ProcessEnv> {
if (!input.executionIds.some(id => vaultExecutionIds.includes(id))) return {};
if (input.executionIds.length !== 1)
throw new Error(
"Vault runtime inputs require the exact single selected source execution",
);
const backend = new URL(input.backendOrigin);
if (
backend.protocol !== "http:" ||
backend.hostname !== "127.0.0.1" ||
backend.origin !== input.backendOrigin
)
throw new Error("Vault runtime inputs require the owned loopback backend");
const filename = input.env[CHAT_PRIVATE_INPUTS_ENV];
if (!filename) return {};
const directory = path.join(input.temporaryRoot, "selected-private-inputs");
if (filename !== path.join(directory, "chat-selected-inputs.json"))
throw new Error(
"Vault runtime input path escaped its owned attempt directory",
);
const uid = process.getuid?.();
if (uid === undefined)
throw new Error("Vault runtime input ownership cannot be verified");
for (const target of [input.temporaryRoot, directory]) {
const info = await lstat(target);
if (
!info.isDirectory() ||
info.isSymbolicLink() ||
!privateVaultMetadataMatches(info, uid)
)
throw new Error("Vault runtime input directory is not private and owned");
}
const handle = await open(
filename,
constants.O_RDONLY | constants.O_NOFOLLOW,
);
let envelope: Envelope;
try {
const before = await handle.stat();
if (
!before.isFile() ||
!privateVaultMetadataMatches(before, uid) ||
before.size > 262144 ||
!(await realpath(filename)).startsWith(
(await realpath(input.temporaryRoot)) + path.sep,
)
)
throw new Error("Vault runtime input file is not private and owned");
const bytes = await handle.readFile(),
after = await handle.stat(),
named = await lstat(filename);
if (
before.ino !== after.ino ||
before.dev !== after.dev ||
before.size !== after.size ||
before.mtimeMs !== after.mtimeMs ||
named.ino !== before.ino ||
named.dev !== before.dev
)
throw new Error("Vault private input changed during read");
try {
envelope = JSON.parse(bytes.toString("utf8")) as Envelope;
} catch {
throw new Error("Vault runtime input JSON is invalid; contents withheld");
}
} finally {
await handle.close();
}
if (envelope.version !== 1 || envelope.executionId !== input.executionIds[0])
throw new Error("Vault runtime input belongs to another execution");
// Ordinary reserved setup handles missing selected provider inputs. It never
// receives an ambient fallback merely because the private binding is absent.
if (!envelope.external?.providers?.vault || envelope.setupBlockers?.length)
return {};
const binding = envelope.external.providers.vault;
if (envelope.external.schemaVersion !== 1 || !validVaultBinding(binding))
throw new Error(
"Vault runtime binding does not designate the exact allowed AWS resources",
);
const declaration = envelope.declaredExternalRequirements;
if (
declaration?.length !== 1 ||
declaration[0]?.provider !== "vault" ||
[...declaration[0].fields].sort().join() !==
[...vaultPrivateFields].sort().join()
)
throw new Error(
"Vault runtime fields differ from the selected fingerprinted declaration",
);
const p = binding.prerequisites;
const read = (
field: "accessKeyIdEnv" | "secretAccessKeyEnv" | "sessionTokenEnv",
) => {
const name = p[field]!;
if (!envelope.declaredReferenceNames?.includes(name))
throw new Error(
"Vault runtime credential is not an explicitly selected reference",
);
const value = envelope.references?.[name];
if (
typeof value !== "string" ||
!value ||
/[\r\n\0]/.test(value) ||
value.length > 16384
)
throw new Error(
"Selected AWS runtime credential is missing or invalid; value withheld",
);
return value;
};
const accessKey = read("accessKeyIdEnv"),
secretKey = read("secretAccessKeyEnv"),
sessionToken = read("sessionTokenEnv");
if (
!/^ASIA[A-Z0-9]{12,124}$/.test(accessKey) ||
secretKey.length < 16 ||
sessionToken.length < 16
)
throw new Error(
"Vault runtime requires explicitly selected temporary AWS credentials",
);
const awsConfig = path.join(directory, "aws-vault-empty-config");
try {
await writeFile(awsConfig, "", { mode: 0o600, flag: "wx" });
} catch (error) {
if ((error as NodeJS.ErrnoException).code !== "EEXIST") throw error;
const info = await lstat(awsConfig);
if (
!info.isFile() ||
info.isSymbolicLink() ||
info.size !== 0 ||
!privateVaultMetadataMatches(info, uid)
)
throw new Error("Selected empty AWS config file is unsafe");
}
return {
AWS_ACCESS_KEY_ID: accessKey,
AWS_SECRET_ACCESS_KEY: secretKey,
AWS_SESSION_TOKEN: sessionToken,
AWS_REGION: p.region,
AWS_DEFAULT_REGION: p.region,
AWS_EC2_METADATA_DISABLED: "true",
AWS_CONFIG_FILE: awsConfig,
AWS_SHARED_CREDENTIALS_FILE: awsConfig,
PAPERCLIP_SECRETS_AWS_REGION: p.region,
PAPERCLIP_SECRETS_AWS_DEPLOYMENT_ID: p.deploymentId,
PAPERCLIP_SECRETS_AWS_KMS_KEY_ID: p.kmsKeyId,
};
}