mirror of
https://github.com/garrytan/gstack.git
synced 2026-08-31 10:20:42 +02:00
* fix(browse): never chmod shared, symlinked, or foreign-owned dirs to 0700 restrictDirectoryPermissions unconditionally chmodded its target. On hosts where the process holds CAP_FOWNER (Docker as root, CI sandboxes) that chmod SUCCEEDS on root-owned /tmp whenever a state file is configured there (BROWSE_STATE_FILE=/tmp/x.json derives stateDir=/tmp), and a 0700 /tmp breaks access(2)-based checks machine-wide for every other process. The POSIX branch now refuses shared sticky dirs, world-writable mounts under root, foreign-owned dirs, and symlinked state dirs; refusals warn once per process instead of failing silent; owned-but-unreadable dirs keep their chmod self-repair; and the check-then-act race is closed with fd-anchored O_NOFOLLOW + fstat/fchmod on a single inode. Regression tests cover the sticky-dir, foreign-uid, mkdirSecure-reapply, and symlinked-dir shapes. * fix: hash with sha256sum before shasum on Linux (config slugs + setup verify) shasum is perl/macOS; coreutils-only Linux ships sha256sum. Two call sites hard-coded shasum: gstack-config's sha8_of/sha16 (so resolve-user-slug exited 127 for any Linux user with a git email, the Layer-3 fallback) and the generated bun-installer checksum snippet in the browse/qa NEEDS_SETUP flow (spurious "checksum mismatch" on the same distros). Both now resolve sha256sum first and fall back to shasum -a 256. New shim-PATH tests pin BOTH hasher branches of sha8_of to a known vector and cover the sha8->sha16 collision escalation end to end. * feat(contract): Aside is the recommended driver for third-party web actions The Third-Party Web Actions contract (ship, spec, office-hours, land-and-deploy, setup-deploy) now names the Aside AI browser as the recommended driver: it acts across the user's real logged-in sessions, which is what vendor-dashboard moments need. Supersedes the v1.65.0.0 de-Aside stance by explicit user directive (2026-08-27). Detection is a runtime probe (command -v + aside --version under a portable gtimeout/timeout/bare guard; nonzero exit = not detected). Consent options render per detection state with Aside recommended and the first-party stack ($B headed + handoff, GStack Browser) as the universal fallback. Absent on macOS, the contract mentions the aside.com download (macOS 15+) once per task; gstack never runs an installer and binary presence is never consent. Drive discipline: step-wise over whole-task delegation, vendor confirm mode on, vendor skill/--help text scoped to operational syntax only, secrets minimized (autofill / human-used copy buttons), Apple credential creation never a drive target in any skill, failure path quotes redacted errors and falls back only with fresh consent. test/third-party-actions.test.ts pins every load-bearing sentence (21 tests) plus repo-wide tripwires: an aside command allowlist (--version/--help only, code spans AND prose) and a ban on Aside installer invocations across all generated docs. Budget ratchet fixture and carve skeleton ceilings refreshed in this commit per the ratchet protocol. * chore: regenerate remaining browse-setup snippet consumers The sha256sum-first checksum fallback in the generated NEEDS_SETUP snippet renders into every browse-consuming skill, not just browse/qa. Mechanical regen of the other ten consumers; no template changes here. * test: consent-gate E2E suite + functional fs-capability probes Five hermetic gate-tier E2E cases (tpa-present / absent-linux / broken / absent-darwin / apple-ban) drive the real contract section through claude -p with PATH shims for aside and uname; the absent cases filter any REAL aside binary out of the child PATH and assert absence with Bun.which before spawning, so dev machines cannot leak into detection. Registered per-case in E2E_TOUCHFILES/E2E_TIERS with template-level deps (ship/SKILL.md.tmpl, gen-skill-docs.ts) and added to the evals.yml matrix with tier: gate. eval:bg:periodic's detach timeout rises to 36000s for the grown periodic shard census (floor-enforced by test/eval-detach-timeout-floor.test.ts); CLAUDE.md doc updated to match. test/helpers/fs-caps.ts adds canRevokeWrites/canRevokeReads functional probes; 13 chmod-based tests swap their uid-0-only guards for the probes so suites skip honestly on CAP_DAC_OVERRIDE containers (this sandbox: uid 1000 with full caps) instead of asserting revocations the kernel ignores. path-validation's symlink test targets /etc/passwd (exists everywhere; /etc/crontab is absent on Amazon Linux). * docs: file the Aside follow-ups in TODOS Phase-2 QA logged-in-evidence path (P3), a hostile-vendor-skill E2E for the contract's override sentence (P2), and fd-anchoring the file-level permission writes to match the directory hardening (P3). * chore: bump version and changelog (v1.72.0.0) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs: update project documentation for v1.72.0.0 docs/skills.md: Third-Party Web Actions subsection under /ship (Aside recommended driver, consent rules, credential boundaries). BROWSER.md: "Aside and third-party drives" subsection under Real-browser mode + ToC entry, including the no-gstack-side-audit-trail caveat (ship adversarial finding 12). TODOS.md: mark the finding-12 doc note done. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs: apply cross-model doc review fixes for v1.72.0.0 docs/skills.md: restore the /ship closing line above the new subsection. BROWSER.md: ToC label matches the heading; BROWSE_STATE_FILE env row documents the new dir-hardening refusal + one-time warning. CHANGELOG: correct the hasher precedence wording (sha256sum first, shasum fallback) and the fs-caps count (14 test files, verified against the diff). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * docs: close cross-model doc-review gaps for v1.72.0.0 setup's manual bun-verify instruction gets the same sha256sum-first fallback the automated snippet got (coreutils-only Linux); BROWSER.md's BROWSE_STATE_FILE row now lists the under-root world-writable refusal; test-cost ceilings in CLAUDE.md/CONTRIBUTING.md updated for the five new gate E2E cases (~$4.20 E2E / ~$4.35 evals). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix(test): gate the symlink-refusal test to POSIX and drop the umask assumption The symlink regression test exercised the POSIX O_NOFOLLOW branch but ran on Windows, where restrictDirectoryPermissions takes the icacls branch and stat has no POSIX modes (0o666 always) — windows-free-tests failed on mode 493 vs 438. Early-return on win32 like every sibling test in the file, and assert the target's mode is UNCHANGED (captured post-mkdir) instead of hardcoding 0o755, which a strict umask would also break. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
285 lines
12 KiB
TypeScript
285 lines
12 KiB
TypeScript
/**
|
|
* User-slug identity resolution chain (T16 / D4 A3).
|
|
*
|
|
* Verifies the gstack-config resolve-user-slug subcommand walks the
|
|
* documented fallback chain:
|
|
* 1. mcp__gbrain__whoami.client_name (skipped when gbrain not on PATH)
|
|
* 2. $USER env var
|
|
* 3. sha8($(git config user.email))
|
|
* 4. anonymous-<sha8(hostname)>
|
|
*
|
|
* Result is persisted under user_slug_at_<endpoint-id> for stability.
|
|
* Test isolation via GSTACK_HOME and HOME env overrides.
|
|
*
|
|
* Gate-tier, free, ~50ms.
|
|
*/
|
|
|
|
import { describe, test, expect, beforeEach, afterEach } from 'bun:test';
|
|
import { mkdtempSync, existsSync, readFileSync, writeFileSync, rmSync, mkdirSync, symlinkSync } from 'fs';
|
|
import { createHash } from 'crypto';
|
|
import { join } from 'path';
|
|
import { tmpdir } from 'os';
|
|
import { spawnSync } from 'child_process';
|
|
|
|
const REPO_ROOT = process.cwd();
|
|
const CONFIG_BIN = join(REPO_ROOT, 'bin', 'gstack-config');
|
|
|
|
let TMP_HOME: string;
|
|
const ORIGINAL = {
|
|
HOME: process.env.HOME,
|
|
GSTACK_HOME: process.env.GSTACK_HOME,
|
|
USER: process.env.USER,
|
|
};
|
|
|
|
function runConfig(args: string[], extraEnv: Record<string, string> = {}): { stdout: string; status: number; stderr: string } {
|
|
const result = spawnSync(CONFIG_BIN, args, {
|
|
encoding: 'utf-8',
|
|
env: {
|
|
...process.env,
|
|
// HOME isolation: endpoint_hash() reads $HOME/.claude.json for the
|
|
// gbrain MCP URL. Pointing HOME at the empty TMP_HOME makes it
|
|
// deterministically 'local' regardless of the developer's real
|
|
// ~/.claude.json (which would otherwise change the persisted key
|
|
// namespace to user_slug_at_<sha8-of-url>).
|
|
HOME: TMP_HOME,
|
|
...extraEnv,
|
|
},
|
|
timeout: 5000,
|
|
});
|
|
return { stdout: result.stdout || '', status: result.status ?? -1, stderr: result.stderr || '' };
|
|
}
|
|
|
|
beforeEach(() => {
|
|
TMP_HOME = mkdtempSync(join(tmpdir(), 'gstack-user-slug-test-'));
|
|
process.env.GSTACK_HOME = TMP_HOME;
|
|
});
|
|
|
|
afterEach(() => {
|
|
for (const [k, v] of Object.entries(ORIGINAL)) {
|
|
if (v !== undefined) process.env[k] = v;
|
|
else delete (process.env as Record<string, unknown>)[k];
|
|
}
|
|
try { rmSync(TMP_HOME, { recursive: true, force: true }); } catch { /* best effort */ }
|
|
});
|
|
|
|
describe('endpoint-hash subcommand', () => {
|
|
test('returns deterministic 8-char hex or literal "local"', () => {
|
|
const result = runConfig(['endpoint-hash'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).toBe(0);
|
|
const out = result.stdout.trim();
|
|
expect(out === 'local' || /^[a-f0-9]{8}$/.test(out) || /^[a-f0-9]{16}$/.test(out)).toBe(true);
|
|
});
|
|
});
|
|
|
|
describe('resolve-user-slug fallback chain', () => {
|
|
test('uses $USER when set (layer 2)', () => {
|
|
const result = runConfig(['resolve-user-slug'], { GSTACK_HOME: TMP_HOME, USER: 'alice-test' });
|
|
expect(result.status).toBe(0);
|
|
expect(result.stdout.trim()).toBe('alice-test');
|
|
});
|
|
|
|
test('lowercases + dash-normalizes $USER', () => {
|
|
const result = runConfig(['resolve-user-slug'], { GSTACK_HOME: TMP_HOME, USER: 'Alice Test' });
|
|
expect(result.status).toBe(0);
|
|
// Spaces become dashes, uppercase becomes lowercase
|
|
expect(result.stdout.trim()).toMatch(/^alice-test$/i);
|
|
});
|
|
|
|
test('falls through past empty $USER to git email or anonymous', () => {
|
|
const result = runConfig(['resolve-user-slug'], { GSTACK_HOME: TMP_HOME, USER: '' });
|
|
expect(result.status).toBe(0);
|
|
const slug = result.stdout.trim();
|
|
expect(slug.length).toBeGreaterThan(0);
|
|
// Should be either email-<sha8> or anonymous-<sha8>
|
|
expect(slug).toMatch(/^(email-|anonymous-)[a-f0-9]+$|^[a-zA-Z0-9-]+$/);
|
|
});
|
|
|
|
test('persists resolution to user_slug_at_<endpoint-id> on first call', () => {
|
|
runConfig(['resolve-user-slug'], { GSTACK_HOME: TMP_HOME, USER: 'persisttest' });
|
|
const configFile = join(TMP_HOME, 'config.yaml');
|
|
expect(existsSync(configFile)).toBe(true);
|
|
const content = readFileSync(configFile, 'utf-8');
|
|
// HOME is isolated to the empty TMP_HOME, so endpoint_hash() is
|
|
// deterministically the literal 'local' on every machine.
|
|
expect(content).toMatch(/^user_slug_at_local:\s+persisttest/m);
|
|
});
|
|
|
|
test('subsequent calls return same slug (stable across sessions)', () => {
|
|
const first = runConfig(['resolve-user-slug'], { GSTACK_HOME: TMP_HOME, USER: 'stabletest' });
|
|
const second = runConfig(['resolve-user-slug'], { GSTACK_HOME: TMP_HOME, USER: 'changed-after' });
|
|
// Second call ignores new $USER because the slug was already persisted.
|
|
expect(first.stdout.trim()).toBe('stabletest');
|
|
expect(second.stdout.trim()).toBe('stabletest');
|
|
});
|
|
});
|
|
|
|
describe('sha8_of portable hash (sha256sum → shasum fallback)', () => {
|
|
// sha8_of must work on coreutils-only Linux (no shasum: the exit-127
|
|
// regression) AND on stock macOS (no sha256sum). The ambient PATH decides
|
|
// which branch runs, so a plain subprocess call only ever covers one branch
|
|
// per platform. Pin BOTH deterministically: extract the real function text
|
|
// from bin/gstack-config (no drift-prone copy) and run it under /bin/sh
|
|
// with a shim-only PATH that makes exactly one hasher visible. The wrong
|
|
// branch exits 127 (its tool is absent from the shim dir), so branch
|
|
// selection is asserted structurally, not inferred.
|
|
const EXPECTED = '2cf24dba'; // sha256("hello") = 2cf24dba5fb0a30e2…
|
|
|
|
function sha8FnSource(): string {
|
|
const src = readFileSync(CONFIG_BIN, 'utf-8');
|
|
const m = src.match(/^sha8_of\(\) \{\n[\s\S]*?\n\}/m);
|
|
if (!m) throw new Error('sha8_of() not found in bin/gstack-config');
|
|
return m[0];
|
|
}
|
|
|
|
/** Absolute-path sha256 pipeline for shims (host has sha256sum OR shasum). */
|
|
function realHasherLine(): string {
|
|
const sha256sum = Bun.which('sha256sum');
|
|
if (sha256sum) return `exec ${sha256sum} "$@"`;
|
|
const shasum = Bun.which('shasum');
|
|
if (shasum) return `exec ${shasum} -a 256 "$@"`;
|
|
throw new Error('neither sha256sum nor shasum available on this host');
|
|
}
|
|
|
|
function runSha8(shimDir: string) {
|
|
const result = spawnSync('/bin/sh', ['-c', `${sha8FnSource()}\nsha8_of "hello"`], {
|
|
encoding: 'utf-8',
|
|
env: { PATH: shimDir }, // ONLY the shim dir: absent tools are really absent
|
|
timeout: 5000,
|
|
});
|
|
return { stdout: (result.stdout || '').trim(), status: result.status ?? -1, stderr: result.stderr || '' };
|
|
}
|
|
|
|
function makeShimDir(): string {
|
|
const dir = mkdtempSync(join(tmpdir(), 'gstack-sha8-shim-'));
|
|
const cut = Bun.which('cut');
|
|
if (!cut) throw new Error('cut not on PATH');
|
|
symlinkSync(cut, join(dir, 'cut'));
|
|
return dir;
|
|
}
|
|
|
|
test('coreutils-only PATH (sha256sum present, shasum absent) — the Linux exit-127 regression', () => {
|
|
const shim = makeShimDir();
|
|
try {
|
|
writeFileSync(join(shim, 'sha256sum'), `#!/bin/sh\n${realHasherLine()}\n`, { mode: 0o755 });
|
|
const result = runSha8(shim);
|
|
expect(result.stderr).toBe('');
|
|
expect(result.status).toBe(0);
|
|
expect(result.stdout).toBe(EXPECTED);
|
|
} finally {
|
|
rmSync(shim, { recursive: true, force: true });
|
|
}
|
|
});
|
|
|
|
test('sha256sum absent falls back to `shasum -a 256` with identical output (macOS branch)', () => {
|
|
const shim = makeShimDir();
|
|
try {
|
|
// Arg-validating shasum shim: wrong/missing `-a 256` exits 64, which
|
|
// would surface as a failed pipeline — pins the exact invocation.
|
|
writeFileSync(
|
|
join(shim, 'shasum'),
|
|
`#!/bin/sh\n[ "$1" = "-a" ] && [ "$2" = "256" ] || exit 64\nshift 2\n${realHasherLine()}\n`,
|
|
{ mode: 0o755 },
|
|
);
|
|
const result = runSha8(shim);
|
|
expect(result.stderr).toBe('');
|
|
expect(result.status).toBe(0);
|
|
expect(result.stdout).toBe(EXPECTED); // same vector ⇒ branches are equivalent
|
|
} finally {
|
|
rmSync(shim, { recursive: true, force: true });
|
|
}
|
|
});
|
|
});
|
|
|
|
describe('endpoint-hash collision escalation (sha8 → sha16)', () => {
|
|
// endpoint_hash_with_collision_check's sha16 arm carries its own portable
|
|
// hash pipeline (sha256sum → shasum). This drives it end-to-end through the
|
|
// real binary: a gbrain MCP URL in $HOME/.claude.json plus config keys at
|
|
// BOTH the sha8 and sha16 namespaces is the recorded-collision evidence
|
|
// that makes `endpoint-hash` emit the 16-char hash. Skipped where jq is
|
|
// absent (the script itself degrades to 'local' there).
|
|
test('emits sha8 normally, sha16 when a stored sha16-namespaced key exists', () => {
|
|
if (!Bun.which('jq')) return; // endpoint_hash requires jq; degrades to 'local' without it
|
|
const url = 'https://gbrain.example.test/mcp';
|
|
const hex = createHash('sha256').update(url).digest('hex');
|
|
const sha8 = hex.slice(0, 8);
|
|
const sha16 = hex.slice(0, 16);
|
|
writeFileSync(join(TMP_HOME, '.claude.json'), JSON.stringify({ mcpServers: { gbrain: { url } } }));
|
|
|
|
// No collision evidence yet → plain sha8.
|
|
const plain = runConfig(['endpoint-hash'], { GSTACK_HOME: TMP_HOME });
|
|
expect(plain.status).toBe(0);
|
|
expect(plain.stdout.trim()).toBe(sha8);
|
|
|
|
// Keys stored at both namespaces → escalate to sha16.
|
|
writeFileSync(
|
|
join(TMP_HOME, 'config.yaml'),
|
|
`brain_trust_policy@${sha8}: personal\nbrain_trust_policy@${sha16}: shared\n`,
|
|
);
|
|
const escalated = runConfig(['endpoint-hash'], { GSTACK_HOME: TMP_HOME });
|
|
expect(escalated.status).toBe(0);
|
|
expect(escalated.stdout.trim()).toBe(sha16);
|
|
});
|
|
});
|
|
|
|
describe('brain_trust_policy@<endpoint-id> namespace', () => {
|
|
test('default value is "unset"', () => {
|
|
const result = runConfig(['get', 'brain_trust_policy@deadbeef'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).toBe(0);
|
|
expect(result.stdout).toBe('unset');
|
|
});
|
|
|
|
test('set + get roundtrip works', () => {
|
|
const setResult = runConfig(['set', 'brain_trust_policy@deadbeef', 'personal'], { GSTACK_HOME: TMP_HOME });
|
|
expect(setResult.status).toBe(0);
|
|
const getResult = runConfig(['get', 'brain_trust_policy@deadbeef'], { GSTACK_HOME: TMP_HOME });
|
|
expect(getResult.stdout).toBe('personal');
|
|
});
|
|
|
|
test('invalid value falls back to unset with warning', () => {
|
|
const result = runConfig(['set', 'brain_trust_policy@deadbeef', 'invalid-value'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).toBe(0);
|
|
expect(result.stderr).toContain('not recognized');
|
|
const getResult = runConfig(['get', 'brain_trust_policy@deadbeef'], { GSTACK_HOME: TMP_HOME });
|
|
expect(getResult.stdout).toBe('unset');
|
|
});
|
|
|
|
test('shared value accepted', () => {
|
|
runConfig(['set', 'brain_trust_policy@deadbeef', 'shared'], { GSTACK_HOME: TMP_HOME });
|
|
const getResult = runConfig(['get', 'brain_trust_policy@deadbeef'], { GSTACK_HOME: TMP_HOME });
|
|
expect(getResult.stdout).toBe('shared');
|
|
});
|
|
|
|
test('per-endpoint policies dont collide', () => {
|
|
runConfig(['set', 'brain_trust_policy@aaaaaaaa', 'personal'], { GSTACK_HOME: TMP_HOME });
|
|
runConfig(['set', 'brain_trust_policy@bbbbbbbb', 'shared'], { GSTACK_HOME: TMP_HOME });
|
|
const a = runConfig(['get', 'brain_trust_policy@aaaaaaaa'], { GSTACK_HOME: TMP_HOME });
|
|
const b = runConfig(['get', 'brain_trust_policy@bbbbbbbb'], { GSTACK_HOME: TMP_HOME });
|
|
expect(a.stdout).toBe('personal');
|
|
expect(b.stdout).toBe('shared');
|
|
});
|
|
});
|
|
|
|
describe('key validation', () => {
|
|
test('rejects keys with disallowed characters', () => {
|
|
const result = runConfig(['get', 'bad-key'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).not.toBe(0);
|
|
expect(result.stderr).toContain('alphanumeric');
|
|
});
|
|
|
|
test('accepts plain alphanumeric/underscore keys', () => {
|
|
const result = runConfig(['get', 'proactive'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).toBe(0);
|
|
});
|
|
|
|
test('accepts @<hex-hash> suffix on key', () => {
|
|
const result = runConfig(['get', 'brain_trust_policy@abc123ff'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).toBe(0);
|
|
});
|
|
|
|
test('accepts @local suffix on key', () => {
|
|
const result = runConfig(['get', 'brain_trust_policy@local'], { GSTACK_HOME: TMP_HOME });
|
|
expect(result.status).toBe(0);
|
|
expect(result.stdout).toBe('unset');
|
|
});
|
|
});
|