fix(browse): remove the dead security shield + unfed /health.security (re-derived from #2557)

The sidebar's SEC shield has been dead UI since the PTY terminal rewrite:
nothing set its data-status, nothing unhid it, and the /health.security
field behind it read getStatus() off ~/.gstack/security/session-state.json
— a file whose ONLY writer (sidebar-agent.ts) was deleted with the chat
path. /health therefore reported a permanent 'inactive', or a stale
FALSE-GREEN 'protected' wherever an old state file survived on disk (a
single unit-test run was enough to plant one). A green shield sourced from
leftover state reads as "no threats detected" when the real state is "not
measured" — the same fail-open class as #2026.

Removed (dead surfaces only): the shield markup/CSS and the stale
sidepanel.js comment; the /health security field and server.ts's getStatus
import; getStatus / SecurityStatus / StatusDetail / SessionState /
read+writeSessionState (and security.ts's dead child_process import); the
session-state + getStatus unit tests — including the round-trip test that
wrote real fixture data into ~/.gstack and left /health green forever.
(The PR's security-sidepanel-dom.test.ts deletion already happened on main
via #2230; its resolveDisconnectCause guard landed via the #2085 typeof
fix. Neither re-applied.)

Kept, per ENG-OV9 — security.ts has LIVE consumers: the pure combiner
(combineVerdict + THRESHOLDS), canary utilities, and extractDomain stay;
server.ts's /pty-inject-scan L4 path (isSidecarAvailable + scanWithSidecar)
is untouched. browse/test/server-security-surface.test.ts pins BOTH
directions: the dead surface stays dead (no /health security field, no
getStatus import, no reader of the security session-state file, shield
markup gone) and the live half stays live (sidecar wiring in server.ts,
combiner/canary exports in security.ts, /health carries no token — the
v1.63 regression wall). A future re-feed from LIVE signals must update
that test deliberately rather than resurrect the state-file path.

F13 (same commit): CLAUDE.md's Sidebar security stack section, ARCHITECTURE.md's
prompt-injection Visibility + critical-constraint paragraphs, and
BROWSER.md's security section now describe the removed surfaces as history,
not live features.

Net -166 lines. Tests: server-security-surface + security +
security-adversarial(+fixes) + security-integration + server-auth 114 pass;
sidepanel-* + extension-token + extension-sender-auth 58 pass / 2 skip.

Re-derived from PR #2557 by @frederik-kaster-noygear.

Co-authored-by: Frederik Kaster <frederik.kaster@noygear.ai>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Garry Tan
2026-08-16 10:59:08 -07:00
co-authored by Frederik Kaster Claude Fable 5
parent af23375aa2
commit f4e84b4dc9
10 changed files with 134 additions and 214 deletions
+13 -94
View File
@@ -23,18 +23,19 @@
* host process): the combiner is pure and tested, and server.ts's
* inline L4 path is the consumer of record.
*
* Cross-process state lives at ~/.gstack/security/session-state.json.
* classifierStatus in that state has no live writer since the chat-path rip
* (the sidecar reports status over its own NDJSON protocol instead).
* There is no longer any cross-process session state (#2557).
* ~/.gstack/security/session-state.json existed to carry classifier status
* across the server.ts / sidebar-agent.ts boundary; sidebar-agent.ts went
* away with the PTY terminal rewrite, leaving nothing to write the file and
* a /health.security status that reported stale or empty data — a permanent
* 'inactive', or a false-green 'protected' wherever an old state file
* survived on disk. getStatus / SessionState / read+writeSessionState and
* the /health field were removed together. Per-tab decision files under
* ~/.gstack/security/decisions/ are unaffected, and the L4 sidecar reports
* status over its own NDJSON protocol (security-sidecar-client.ts).
*/
import { randomBytes, createHash } from 'crypto';
import { spawn } from 'child_process';
import * as fs from 'fs';
import * as path from 'path';
import * as os from 'os';
import { restrictFilePermissions, appendSecureFile, mkdirSecure } from './file-permissions';
import { atomicWriteQuiet } from '../../lib/fs-atomic';
import { randomBytes } from 'crypto';
// ─── Thresholds + verdict types ──────────────────────────────
@@ -83,17 +84,6 @@ export interface SecurityResult {
confidence: number;
}
export type SecurityStatus = 'protected' | 'degraded' | 'inactive';
export interface StatusDetail {
status: SecurityStatus;
layers: {
testsavant: 'ok' | 'degraded' | 'off';
canary: 'ok' | 'off';
};
lastUpdated: string;
}
// ─── Verdict combiner (ensemble rule, label-first for transcript) ────
/**
@@ -322,79 +312,8 @@ export function checkCanaryInStructure(value: unknown, canary: string): boolean
// attempts.jsonl rotation + telemetry spawn plumbing) lived here until the
// chat-path scanner that called it was ripped with sidebar-agent.ts. The
// LIVE attempts.jsonl writer is tunnel-denial-log.ts, which owns its own
// rotation.
const SECURITY_DIR = path.join(os.homedir(), '.gstack', 'security');
// ─── Cross-process session state ─────────────────────────────
const STATE_FILE = path.join(SECURITY_DIR, 'session-state.json');
/**
* SessionState is a DISK FORMAT (~/.gstack/security/session-state.json).
* Old files may carry a `transcript` field inside classifierStatus from the
* removed Haiku layer — readSessionState tolerates it (JSON.parse keeps the
* extra key; getStatus ignores it), but we never write it.
*/
export interface SessionState {
sessionId: string;
canary: string;
warnedDomains: string[]; // per-session rate limit for special telemetry
classifierStatus: {
testsavant: 'ok' | 'degraded' | 'off';
};
lastUpdated: string;
}
/**
* Atomic write of session state (via lib/fs-atomic). Writes are safe
* across process boundaries. Swallow-with-log polarity: a failed write
* must never take down the caller (security state is best-effort cache).
*/
export function writeSessionState(state: SessionState): void {
try { mkdirSecure(SECURITY_DIR); } catch { /* write below fails and logs */ }
if (atomicWriteQuiet(STATE_FILE, JSON.stringify(state, null, 2), { mode: 0o600 })) {
// Windows ACL hardening (POSIX chmod is redundant with mode above).
restrictFilePermissions(STATE_FILE);
} else {
console.error('[security] writeSessionState failed');
}
}
export function readSessionState(): SessionState | null {
try {
if (!fs.existsSync(STATE_FILE)) return null;
return JSON.parse(fs.readFileSync(STATE_FILE, 'utf8'));
} catch {
return null;
}
}
// ─── Status reporting (for shield icon via /health) ──────────
export function getStatus(): StatusDetail {
const state = readSessionState();
// Read the field explicitly (never spread classifierStatus): old on-disk
// state may carry a stale `transcript` key from the removed Haiku layer,
// and spreading would leak it into the /health payload.
const testsavant = state?.classifierStatus?.testsavant ?? 'off';
const canary = state?.canary ? 'ok' : 'off';
let status: SecurityStatus;
if (testsavant === 'ok' && canary === 'ok') {
status = 'protected';
} else if (testsavant === 'off' && canary === 'off') {
status = 'inactive';
} else {
status = 'degraded';
}
return {
status,
layers: { testsavant, canary: canary as 'ok' | 'off' },
lastUpdated: state?.lastUpdated ?? new Date().toISOString(),
};
}
// rotation. The cross-process session state + getStatus shield feed went
// the same way (#2557) — see the module header.
/**
* Extract url domain for logging. Never logs path or query string.
+7 -5
View File
@@ -24,7 +24,6 @@ import {
runContentFilters, type ContentFilterResult,
markHiddenElements, getCleanTextWithStripping, cleanupHiddenMarkers,
} from './content-security';
import { getStatus as getSecurityStatus } from './security';
import { isSidecarAvailable, scanWithSidecar } from './security-sidecar-client';
import { writeSecureFile, mkdirSecure, appendSecureFile } from './file-permissions';
import { handleSnapshot, SNAPSHOT_FLAGS } from './snapshot';
@@ -1899,10 +1898,13 @@ export function buildFetchHandler(cfg: ServerConfig): ServerHandle {
mode: browserManager.getConnectionMode(),
uptime: Math.floor((Date.now() - startTime) / 1000),
tabs: browserManager.getTabCount(),
// Security module status — drives the shield icon in the sidepanel.
// Returns {status: 'protected'|'degraded'|'inactive', layers: {...}}.
// Fed by the page-content side (testsavant sidecar, canary state).
security: getSecurityStatus(),
// No `security` field (#2557): the only writer of the status it
// reported (sidebar-agent.ts's session-state file) went away with
// the chat path, so it read from a file nothing wrote — reporting
// a permanent 'inactive', or a stale false-green 'protected'
// wherever an old state file survived on disk. The live defenses
// (content-security L1-L3, the L4 sidecar on /pty-inject-scan)
// report through their own call sites, not through /health.
// Terminal-agent discovery. ONLY a port number — never a token.
// Tokens flow via the /pty-session HttpOnly cookie path. See
// `pty-session-cookie.ts` for the rationale (codex outside-voice
+7 -57
View File
@@ -10,18 +10,12 @@
*/
import { describe, test, expect } from 'bun:test';
import * as fs from 'fs';
import * as os from 'os';
import * as path from 'path';
import {
THRESHOLDS,
combineVerdict,
generateCanary,
injectCanary,
checkCanaryInStructure,
writeSessionState,
readSessionState,
getStatus,
extractDomain,
type LayerSignal,
} from '../src/security';
@@ -244,57 +238,13 @@ describe('canary', () => {
// ─── Attack log + rotation ───────────────────────────────────
// ─── Session state (cross-process, atomic) ───────────────────
describe('session state', () => {
test('write + read round-trip', () => {
const state = {
sessionId: 'test-session-123',
canary: 'CANARY-TEST',
warnedDomains: ['example.com'],
classifierStatus: { testsavant: 'ok' as const },
lastUpdated: '2026-04-19T12:34:56Z',
};
writeSessionState(state);
const got = readSessionState();
expect(got).not.toBeNull();
expect(got!.sessionId).toBe('test-session-123');
expect(got!.canary).toBe('CANARY-TEST');
expect(got!.warnedDomains).toEqual(['example.com']);
});
test('tolerates stale transcript field from pre-rip on-disk state', () => {
// SessionState is a disk format. Files written before the Haiku
// transcript layer was removed carry classifierStatus.transcript —
// getStatus must read them fine, not require transcript for
// 'protected', and never leak the stale key into /health.
const stateFile = path.join(os.homedir(), '.gstack', 'security', 'session-state.json');
fs.mkdirSync(path.dirname(stateFile), { recursive: true });
fs.writeFileSync(stateFile, JSON.stringify({
sessionId: 'legacy-session',
canary: 'CANARY-LEGACY',
warnedDomains: [],
classifierStatus: { testsavant: 'ok', transcript: 'degraded' },
lastUpdated: '2026-04-19T12:34:56Z',
}));
const s = getStatus();
expect(s.status).toBe('protected');
expect('transcript' in s.layers).toBe(false);
});
});
// ─── Status reporting for shield icon ────────────────────────
describe('getStatus', () => {
test('returns a valid SecurityStatus shape', () => {
const s = getStatus();
expect(['protected', 'degraded', 'inactive']).toContain(s.status);
expect(s.layers).toBeDefined();
expect(['ok', 'degraded', 'off']).toContain(s.layers.testsavant);
expect(['ok', 'off']).toContain(s.layers.canary);
expect(s.lastUpdated).toBeTruthy();
});
});
// NOTE (#2557): the session-state + getStatus tests that lived here wrote
// REAL fixture data into ~/.gstack/security/session-state.json — after which
// /health reported a false-green 'protected' indefinitely. The surfaces they
// covered (SessionState, read/writeSessionState, getStatus, the /health
// security field, the sidepanel SEC shield) were dead since the PTY terminal
// rewrite and are now removed. server-security-surface.test.ts pins the
// removal + the live L4 wiring.
// ─── URL domain extraction ───────────────────────────────────
@@ -0,0 +1,86 @@
/**
* #2557 / ENG-OV9: pins the dead-shield removal AND the live L4 wiring.
*
* The removed surface: /health's `security` field read getStatus(), whose
* only data source (~/.gstack/security/session-state.json) lost its only
* writer when sidebar-agent.ts was ripped — so /health reported a permanent
* 'inactive' or, wherever an old state file survived, a stale FALSE-GREEN
* 'protected' ("no threats detected" when the real state was "not
* measured"). Same fail-open class as #2026.
*
* The kept surface (ENG-OV9): security.ts is NOT dead — server.ts's
* /pty-inject-scan path is the live L4 consumer (sidecar scan + URL
* blocklist + datamark envelope), and security.ts's pure combiner/canary
* exports stay. This test pins both directions so a future "cleanup" can't
* silently take the live half, and a future re-feed of /health.security
* from LIVE signals (isSidecarAvailable, content filters) must update this
* test deliberately rather than resurrect the state-file path.
*
* Source-level, same style as windows-spawn-hide.test.ts.
*/
import { describe, expect, test } from 'bun:test';
import * as fs from 'fs';
import * as path from 'path';
const SRC = (f: string) => fs.readFileSync(path.join(import.meta.dir, '../src', f), 'utf-8');
describe('#2557: dead shield surface stays dead', () => {
test('/health carries no security field and server.ts does not import getStatus', () => {
const server = SRC('server.ts');
expect(server).not.toMatch(/security:\s*getSecurityStatus\(\)/);
expect(server).not.toMatch(/getStatus as getSecurityStatus/);
// The SECURITY session-state file must not be read anywhere in src/ —
// that file has no writer, so any reader is a false-signal feed.
// (session-persist.ts's per-project <stateDir>/session-state.json is a
// different, live file — only the ~/.gstack/security/ one is dead.)
for (const f of fs.readdirSync(path.join(import.meta.dir, '../src')).filter((x) => x.endsWith('.ts'))) {
const code = SRC(f).replace(/\/\*[\s\S]*?\*\//g, '').replace(/^\s*\/\/.*$/gm, '').replace(/^\s*\*.*$/gm, '');
const refs = /security[/'",\s][^\n]{0,80}session-state\.json/.test(code);
expect({ file: f, refs }).toEqual({ file: f, refs: false });
}
});
test('security.ts no longer exports the unfed status surface', () => {
const security = SRC('security.ts');
expect(security).not.toMatch(/export function getStatus/);
expect(security).not.toMatch(/export function (read|write)SessionState/);
expect(security).not.toMatch(/export interface SessionState/);
expect(security).not.toMatch(/export interface StatusDetail/);
});
test('the sidepanel shield markup is gone', () => {
const html = fs.readFileSync(path.join(import.meta.dir, '../../extension/sidepanel.html'), 'utf-8');
const css = fs.readFileSync(path.join(import.meta.dir, '../../extension/sidepanel.css'), 'utf-8');
expect(html).not.toContain('security-shield');
expect(css).not.toMatch(/\.security-shield\s*\{/);
});
});
describe('ENG-OV9: the LIVE L4 path is untouched', () => {
test('server.ts still consumes the sidecar on the inject-scan path', () => {
const server = SRC('server.ts');
expect(server).toContain("from './security-sidecar-client'");
expect(server).toMatch(/isSidecarAvailable/);
expect(server).toMatch(/scanWithSidecar\(/);
});
test('security.ts keeps the pure combiner + canary exports', () => {
const security = SRC('security.ts');
expect(security).toMatch(/export const THRESHOLDS/);
expect(security).toMatch(/export function combineVerdict/);
expect(security).toMatch(/export function generateCanary/);
expect(security).toMatch(/export function injectCanary/);
expect(security).toMatch(/export function checkCanaryInStructure/);
expect(security).toMatch(/export function extractDomain/);
});
test('/health stays liveness-only: no token in any mode (regression wall from v1.63)', () => {
const server = SRC('server.ts');
// The /health handler block must not interpolate a token.
const healthIdx = server.indexOf("url.pathname === '/health'");
expect(healthIdx).toBeGreaterThan(0);
const healthBlock = server.slice(healthIdx, healthIdx + 1500);
expect(healthBlock).not.toMatch(/token:\s*[^n]/i);
});
});