mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-08 11:23:40 +01:00
## Thinking Path > - Paperclip uses real Product E2E journeys to verify agent work. > - The trusted TestOps controller must bind the reviewed source and retain inspectable evidence. > - The last uniform attempt exposed bounded observation and evidence-retention defects. Its original outcomes remain unchanged. > - The reviewed successor corrects those boundaries with the original model, requests, counts, and clocks. > - Its private framework report stores canonical attachments once. The controller must validate and restore those references before normal reporting. > - This PR ports that reader support and pins the qualified source in one infrastructure change. It does not merge the full catalog or its core repair PRs. ## Linked Issues or Issue Description Refs #19127, #19083, #19086, #19084, #19093, #19095, #19100, #19112. **What happened?** The last uniform attempt retained real failed outcomes. It also lost the multipart-question original when evidence packaging exceeded its unchanged physical byte allowance. Review found independent boundaries in skill navigation, announcement observation, timeline attribution, hire history, credential questions, and team visibility. **Expected behavior** The next run must execute the same ten requests against one reviewed source. The controller must retain complete original evidence, validate private attachment references, and report the actual outcome. Historical failures and missing-original UNKNOWN grades must remain unchanged. **Steps to reproduce** Use the focused source controls linked above. The private framework control uses a real failed Playwright test and a 140 MiB attachment. Legacy duplicate carriers hit the old cap. Canonical original evidence passes both seals and strict archive restoration. The controller restores derived files and the existing TestOps exporter preserves the failed result. **Paperclip version/commit** Qualified candidate: `155dd1bc1c4b85ae107f6a0cd429720caa281367`. The controller port is based on current master. The full candidate catalog stays on its own immutable branch. **Deployment mode** Existing AWS/TestOps job 100, plan 102, branch master. This PR does not dispatch a run. ## What Changed - Pin the published qualified source and its exact launcher and descriptor fixtures. - Preserve current manifest history. Update twelve declaration fields across eleven cases: oracle versions and skill source-file provenance only. Preserve all 244 identities, 242 canonical declarations, queries, setup, human/native counts, clocks, profile, model, retry policy, and global cap of 8. - Port the exact source-owned compact framework helper and private-transfer reader. Keep the three existing required source/controller protocol helpers byte-identical. - Validate both sealed inventories before restoring registered attachments into the derived selected report tree. Verify exact bytes and hashes. Preserve seals and unregistered evidence. - Keep deep JSON, decoded-body, sensitive-key, path, and original/final status checks. Plain-text bodies and registered plain files fail closed if existing diagnostic redaction would change them; JSON remains under structured inspection. Keep all existing physical byte, entry, ZIP, time, privacy, and archive bounds. - Preserve registered clean metadata and attachment bytes. Reject content that needs privacy redaction instead of silently rewriting immutable evidence. ## Verification - 131 existing Node pin, selection, admission, adapter, reconciliation, and security controls pass. The explicit new target assertion and drift negatives remain active. - 86 current-controller controls pass. They cover legacy reports, privacy and static assets, canonical reference guards, both seals, strict archive restoration, derived files, the actual report normalizer, and the actual TestOps/Allure exporter. The original failed result stays failed; screenshot and trace hashes resolve. - Actual P1 controls reject both an unknown password in a decoded plain-text body and an unknown command token in a registered text file. Parent code fails both controls; corrected code passes. The composed source passes 87 affected controls, fresh server declarations, and full E2E typechecking. The old cfe head completed all 49 CI jobs without adverse conclusions, but its review found this real privacy defect. That CI result is not credited to this successor. - The existing report-lock preflight returns already-reviewed at `60f4c06bca47af3ab9d74281f2ef2e1ff559d566b862ba51c2fbe6c8e5bac35b`. - The existing source SDK guard confirms 506 registered inputs, unchanged registry `efa70621e9fb42aa185f7ee616a1a726aa3eccb367e67e28817cedf26ef2ade8`, and unchanged runtime pins. No registry renewal or SDK build was needed. - Source launcher SHA-256: `2b9a6953ee51f0ea0dbbd475b0491d9a6c39a1d6835eeaa912e0dbb7cf987724`. Descriptor SHA-256: `e026d5c000102f8d50397dc59055d187388304f91dbca9a6ccfbe747004cd16d`. - No model call, evaluation dispatch, evidence download, reporting retry, core merge, or permission change was performed. Full current-head CI and review are pending. Separate source PR failures remain disclosed; local controls do not establish universal source CI readiness. - Seven cases have verified passing historical attempts. A uniform passing ten-case run remains unverified. The old lost multipart-question original remains UNKNOWN. ## Risks This changes a trusted evidence reader and the source selected for future measurements. The new helper is included in the controller import closure. Compact references require matching originals; missing, foreign, changed, unsafe, or redaction-required evidence fails closed. Legacy reporter formats remain valid. A single canonical evidence set can still exceed a bound. Current-head CI and material review must pass before normal protected merge. Actual AWS/TestOps evidence remains required afterward. ## Model Used OpenAI `gpt-6.1-sol`, with reasoning, code editing, and tool use. The host does not expose the context-window size. ## 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
174 lines
12 KiB
TypeScript
174 lines
12 KiB
TypeScript
import { constants } from "node:fs";
|
|
import { createRequire } from "node:module";
|
|
import { createHash } from "node:crypto";
|
|
import { copyFile, lstat, mkdir, readFile, realpath, rm, writeFile } from "node:fs/promises";
|
|
import path from "node:path";
|
|
import { assertSecretFree, findSecretLeak, findSecretLeakInJsonValues, normalizedSecrets, sanitizeJson, redactText } from "./redaction.js";
|
|
|
|
export const PRIVATE_FRAMEWORK_ORIGINAL_REPORT = "snapshots/playwright-framework-original.txt";
|
|
export const PRIVATE_FRAMEWORK_REPORT = "snapshots/playwright-framework.json";
|
|
const producerVersion: string = createRequire(import.meta.url)("@playwright/test/package.json").version;
|
|
const LIMIT = 256 * 1024 * 1024;
|
|
const hash = (data: Buffer) => createHash("sha256").update(data).digest("hex");
|
|
const fail = (): never => { throw new Error("Invalid private framework evidence"); };
|
|
type Row = { path: string; bytes: number; sha256: string };
|
|
type FrameworkReport = { schema: string; sourceReportSha256: string; report: Record<string, unknown>; attachments: Row[]; canonicalExecutionId?: string };
|
|
async function regular(root: string, relative: string, limit = LIMIT) {
|
|
if (!relative || relative.includes("\\") || relative.split("/").some(p => !p || p === "." || p === "..") || /[\x00-\x1f:]/.test(relative)) fail();
|
|
let current = root;
|
|
for (const part of relative.split("/")) { current = path.join(current, part); const entry = await lstat(current); if (entry.isSymbolicLink()) fail(); }
|
|
const info = await lstat(current);
|
|
if (!info.isFile() || info.size > limit || !(await realpath(current)).startsWith(`${await realpath(root)}${path.sep}`)) fail();
|
|
const data = await readFile(current), after = await lstat(current);
|
|
if (data.length !== info.size || after.ino !== info.ino || after.mtimeMs !== info.mtimeMs || after.ctimeMs !== info.ctimeMs) fail();
|
|
return data;
|
|
}
|
|
function results(report: Record<string, unknown>) {
|
|
if (!report.config || typeof report.config !== "object" || !Array.isArray(report.suites) || !report.suites.length || !Array.isArray(report.errors)) fail();
|
|
if ((report.config as Record<string, unknown>).version !== producerVersion) fail();
|
|
const found: Array<Record<string, unknown>> = []; let visited = 0;
|
|
const suite = (value: any) => {
|
|
if (++visited > 4096 || !value || (value.specs !== undefined && !Array.isArray(value.specs)) || (value.suites !== undefined && !Array.isArray(value.suites))) fail();
|
|
for (const spec of value.specs ?? []) {
|
|
if (!spec || !Array.isArray(spec.tests) || !spec.tests.length) fail();
|
|
for (const test of spec.tests) {
|
|
if (!test || !Array.isArray(test.results) || !test.results.length) fail();
|
|
for (const result of test.results) {
|
|
if (++visited > 4096 || !result || !["passed", "failed", "timedOut", "skipped", "interrupted"].includes(result.status)
|
|
|| !Array.isArray(result.attachments) || !Array.isArray(result.errors) || !Array.isArray(result.stdout) || !Array.isArray(result.stderr)) fail();
|
|
found.push(result);
|
|
}
|
|
}
|
|
}
|
|
for (const child of value.suites ?? []) suite(child);
|
|
};
|
|
for (const value of report.suites as unknown[]) suite(value);
|
|
if (!found.length) fail();
|
|
return found;
|
|
}
|
|
export function inspectPrivateFrameworkMetadata(relative: string, raw: string, secrets: readonly string[]) {
|
|
const original = relative === PRIVATE_FRAMEWORK_ORIGINAL_REPORT || relative.endsWith(`/${PRIVATE_FRAMEWORK_ORIGINAL_REPORT}`) || relative === "framework-report.json";
|
|
const sidecar = relative === PRIVATE_FRAMEWORK_REPORT || relative.endsWith(`/${PRIVATE_FRAMEWORK_REPORT}`);
|
|
if (!original && !sidecar) return false;
|
|
if (Buffer.byteLength(raw) > 128 * 1024 * 1024) fail();
|
|
const parsed = JSON.parse(raw), report = original ? parsed : parsed.report;
|
|
const expanded = normalizedSecrets(secrets).flatMap(v => [v, encodeURIComponent(v), Buffer.from(v).toString("base64"), Buffer.from(v).toString("base64url")]);
|
|
if (findSecretLeakInJsonValues(parsed, expanded) || JSON.stringify(sanitizeJson(parsed, secrets)) !== JSON.stringify(parsed)) fail();
|
|
for (const result of results(report)) for (const attachment of result.attachments as any[]) {
|
|
if (!attachment || typeof attachment.name !== "string" || typeof attachment.contentType !== "string") fail();
|
|
if (attachment.body !== undefined) {
|
|
if (typeof attachment.body !== "string" || !/^(?:[A-Za-z0-9+/]{4})*(?:[A-Za-z0-9+/]{2}==|[A-Za-z0-9+/]{3}=)?$/.test(attachment.body)) fail();
|
|
const body = Buffer.from(attachment.body, "base64"), text = body.toString("utf8");
|
|
if (!["text/plain", "application/json"].includes(attachment.contentType) || !Buffer.from(text).equals(body) || findSecretLeak(body, expanded)) fail();
|
|
if (attachment.contentType === "text/plain" && redactText(text, secrets) !== text) fail();
|
|
if (attachment.contentType === "application/json" && (findSecretLeak(JSON.stringify(JSON.parse(text)), expanded) || JSON.stringify(sanitizeJson(JSON.parse(text), secrets)) !== JSON.stringify(JSON.parse(text)))) fail();
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
export async function copyPrivateFrameworkReport(source: string, destination: string, secrets: readonly string[]) {
|
|
const raw = await regular(source, "framework-report.json", 128 * 1024 * 1024);
|
|
if (!Buffer.from(raw.toString("utf8")).equals(raw)) fail();
|
|
assertSecretFree(raw.toString("utf8"), secrets, "private framework metadata");
|
|
const report = JSON.parse(raw.toString("utf8"));
|
|
const attachments = new Map<string, Row>();
|
|
inspectPrivateFrameworkMetadata("framework-report.json", raw.toString("utf8"), secrets);
|
|
for (const result of results(report)) for (const attachment of result.attachments as any[]) {
|
|
if (!attachment || typeof attachment.name !== "string" || typeof attachment.contentType !== "string") fail();
|
|
if (attachment.path === undefined) { if (attachment.body === undefined) fail(); continue; }
|
|
if (typeof attachment.path !== "string" || !path.isAbsolute(attachment.path)) fail();
|
|
const absolute = path.resolve(attachment.path), output = path.join(source, "playwright-output");
|
|
let relative: string;
|
|
if (absolute.startsWith(`${path.resolve(destination)}${path.sep}`)) relative = path.relative(destination, absolute);
|
|
else if (absolute.startsWith(`${path.resolve(output)}${path.sep}`)) relative = path.relative(source, absolute);
|
|
else fail();
|
|
relative = relative!.split(path.sep).join("/");
|
|
const data = await regular(absolute.startsWith(`${path.resolve(destination)}${path.sep}`) ? destination : source, relative);
|
|
if (relative === "result.json") {
|
|
relative = "snapshots/playwright-attached-result.json";
|
|
await mkdir(path.dirname(path.join(destination, relative)), { recursive: true, mode: 0o700 });
|
|
await writeFile(path.join(destination, relative), data, { flag: "wx", mode: 0o600 });
|
|
} else {
|
|
const copied = await regular(destination, relative);
|
|
if (!copied.equals(data)) fail();
|
|
}
|
|
attachment.path = relative;
|
|
const row = { path: relative, bytes: data.length, sha256: hash(data) };
|
|
const previous = attachments.get(relative); if (previous && JSON.stringify(previous) !== JSON.stringify(row)) fail();
|
|
attachments.set(relative, row);
|
|
}
|
|
const sidecar: FrameworkReport = { schema: "paperclip.private-framework-json/v1", sourceReportSha256: hash(raw), report,
|
|
attachments: [...attachments.values()].sort((a, b) => a.path.localeCompare(b.path)) };
|
|
await mkdir(path.join(destination, "snapshots"), { recursive: true, mode: 0o700 });
|
|
await writeFile(path.join(destination, PRIVATE_FRAMEWORK_ORIGINAL_REPORT), raw, { flag: "wx", mode: 0o600 });
|
|
await writeFile(path.join(destination, PRIVATE_FRAMEWORK_REPORT), `${JSON.stringify(sidecar)}\n`, { flag: "wx", mode: 0o600 });
|
|
}
|
|
export async function validatePrivateFrameworkReport(directory: string, canonical = directory, executionId?: string) {
|
|
const raw = await regular(directory, PRIVATE_FRAMEWORK_REPORT, 128 * 1024 * 1024);
|
|
const value = JSON.parse(raw.toString("utf8")) as FrameworkReport;
|
|
const keys = Object.keys(value).sort().join();
|
|
if (keys !== (executionId ? "attachments,canonicalExecutionId,report,schema,sourceReportSha256" : "attachments,report,schema,sourceReportSha256")
|
|
|| value.schema !== "paperclip.private-framework-json/v1" || !/^[0-9a-f]{64}$/.test(value.sourceReportSha256)
|
|
|| executionId && value.canonicalExecutionId !== executionId || !Array.isArray(value.attachments) || value.attachments.length > 4096) fail();
|
|
if (executionId) {
|
|
const original = await validatePrivateFrameworkReport(canonical);
|
|
const { canonicalExecutionId: _, ...retained } = value;
|
|
if (JSON.stringify(retained) !== JSON.stringify(original)) fail();
|
|
}
|
|
const originalRaw = await regular(directory, PRIVATE_FRAMEWORK_ORIGINAL_REPORT, 128 * 1024 * 1024);
|
|
if (hash(originalRaw) !== value.sourceReportSha256) fail();
|
|
inspectPrivateFrameworkMetadata(PRIVATE_FRAMEWORK_ORIGINAL_REPORT, originalRaw.toString("utf8"), []);
|
|
const authoritative = JSON.parse(originalRaw.toString("utf8"));
|
|
const originalResults = results(authoritative), retainedResults = results(value.report);
|
|
if (originalResults.length !== retainedResults.length) fail();
|
|
for (let i = 0; i < originalResults.length; i++) {
|
|
const originalAttachments = originalResults[i]!.attachments as any[], retainedAttachments = retainedResults[i]!.attachments as any[];
|
|
if (originalAttachments.length !== retainedAttachments.length) fail();
|
|
for (let j = 0; j < originalAttachments.length; j++) {
|
|
if (originalAttachments[j].path !== undefined) originalAttachments[j].path = retainedAttachments[j].path;
|
|
}
|
|
}
|
|
if (JSON.stringify(authoritative) !== JSON.stringify(value.report)) fail();
|
|
const referenced = new Set<string>();
|
|
for (const result of results(value.report)) for (const attachment of result.attachments as any[]) {
|
|
if (attachment.path !== undefined) { if (typeof attachment.path !== "string") fail(); referenced.add(attachment.path); }
|
|
}
|
|
const seen = new Set<string>(); let attachmentBytes = 0;
|
|
for (const row of value.attachments) {
|
|
if (!row || Object.keys(row).sort().join() !== "bytes,path,sha256" || !referenced.has(row.path) || seen.has(row.path)
|
|
|| !Number.isSafeInteger(row.bytes) || row.bytes < 0 || !/^[0-9a-f]{64}$/.test(row.sha256)) fail();
|
|
attachmentBytes += row.bytes; if (attachmentBytes > LIMIT) fail();
|
|
const data = await regular(canonical, row.path); if (data.length !== row.bytes || hash(data) !== row.sha256) fail();
|
|
seen.add(row.path);
|
|
}
|
|
if (seen.size !== referenced.size) fail();
|
|
return value;
|
|
}
|
|
export async function referenceSealedFrameworkAttachments(directory: string, original: string, executionId: string) {
|
|
const value = await validatePrivateFrameworkReport(directory);
|
|
const sealed = await validatePrivateFrameworkReport(original);
|
|
if (JSON.stringify(value) !== JSON.stringify(sealed)) fail();
|
|
for (const row of value.attachments.filter(r => r.path.startsWith("playwright-output/"))) {
|
|
const current = await regular(directory, row.path), canonical = await regular(original, row.path);
|
|
if (!current.equals(canonical)) fail();
|
|
await rm(path.join(directory, row.path));
|
|
}
|
|
value.canonicalExecutionId = executionId;
|
|
await writeFile(path.join(directory, PRIVATE_FRAMEWORK_REPORT), `${JSON.stringify(value)}\n`, { mode: 0o600 });
|
|
}
|
|
|
|
export async function materializePrivateFrameworkAttachments(directory: string, original: string, executionId: string) {
|
|
const value = await validatePrivateFrameworkReport(directory, original, executionId);
|
|
let bytes = 0;
|
|
for (const row of value.attachments) {
|
|
bytes += row.bytes; if (bytes > LIMIT) fail();
|
|
if (!row.path.startsWith("playwright-output/")) continue;
|
|
const source = await regular(original, row.path);
|
|
if (source.length !== row.bytes || hash(source) !== row.sha256) fail();
|
|
await mkdir(path.dirname(path.join(directory, row.path)), { recursive: true, mode: 0o700 });
|
|
await copyFile(path.join(original, row.path), path.join(directory, row.path), constants.COPYFILE_EXCL);
|
|
const restored = await regular(directory, row.path);
|
|
if (hash(restored) !== row.sha256) fail();
|
|
}
|
|
}
|