Files
gstack/test/gbrain-repo-policy-client.test.ts
T
Garry TanandClaude Fable 5 b5a951e623 v1.72.0.0 feat: Aside recommended driver for third-party web actions (#2710)
* 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>
2026-08-28 11:43:13 -07:00

212 lines
9.2 KiB
TypeScript

/**
* lib/gbrain-repo-policy-client — batch trust-tier lookup (#2392).
*
* Covers the `get --batch` verb of bin/gstack-gbrain-repo-policy (bash level:
* multiple urls in, per-line verdicts out, input order preserved) and the
* TypeScript client `repoPolicyTierBatch` (ONE spawn, dedup, fast paths, and
* the whole-batch `unreadable` classification on a corrupt store).
*
* Each test uses a temp GSTACK_HOME so nothing leaks into the user's real
* ~/.gstack. The client is exercised against the REAL bash script — the
* script owns URL normalization, so stub stores are seeded through its own
* `set` verb.
*/
import { describe, test, expect, beforeEach, afterEach } from "bun:test";
import { canRevokeReads } from "./helpers/fs-caps";
import * as fs from "fs";
import * as path from "path";
import * as os from "os";
import { spawnSync } from "child_process";
import { repoPolicyTierBatch } from "../lib/gbrain-repo-policy-client";
import { canonicalizeRemote } from "../lib/gstack-memory-helpers";
const ROOT = path.resolve(import.meta.dir, "..");
const BIN = path.join(ROOT, "bin", "gstack-gbrain-repo-policy");
let tmpHome: string;
function env(): NodeJS.ProcessEnv {
return { ...process.env, GSTACK_HOME: tmpHome };
}
function run(args: string[], input?: string) {
const res = spawnSync(BIN, args, { env: env(), encoding: "utf-8", input });
return {
stdout: res.stdout || "",
stderr: res.stderr || "",
status: res.status ?? -1,
};
}
function policyFile(): string {
return path.join(tmpHome, "gbrain-repo-policy.json");
}
beforeEach(() => {
tmpHome = fs.mkdtempSync(path.join(os.tmpdir(), "gbrain-policy-client-"));
});
afterEach(() => {
fs.rmSync(tmpHome, { recursive: true, force: true });
});
describe("bin/gstack-gbrain-repo-policy get --batch (bash level)", () => {
test("multiple urls in, per-line verdicts out, input order preserved", () => {
expect(run(["set", "https://github.com/foo/bar.git", "deny"]).status).toBe(0);
expect(run(["set", "git@github.com:baz/qux.git", "read-only"]).status).toBe(0);
expect(run(["set", "https://github.com/rw/repo", "read-write"]).status).toBe(0);
const r = run(
["get", "--batch"],
// Mixed URL forms — the script's normalize() collapses them to the
// stored keys. `nope/never` has no entry → none.
"git@github.com:foo/bar.git\nhttps://github.com/nope/never\nhttps://github.com/baz/qux\nhttps://github.com/rw/repo.git\n",
);
expect(r.status).toBe(0);
expect(r.stdout).toBe("deny\nnone\nread-only\nread-write\n");
});
test("no store on disk: every line is none, and no file is created", () => {
const r = run(["get", "--batch"], "https://github.com/a/a\nhttps://github.com/b/b\n");
expect(r.status).toBe(0);
expect(r.stdout).toBe("none\nnone\n");
expect(fs.existsSync(policyFile())).toBe(false);
});
test("corrupt store: hard error (exit 2), NOT quarantined, names recovery", () => {
fs.writeFileSync(policyFile(), "not valid json{", { mode: 0o600 });
const r = run(["get", "--batch"], "https://github.com/foo/bar\n");
expect(r.status).toBe(2);
expect(r.stderr).toContain("corrupt");
expect(r.stderr).toContain("gstack-gbrain-repo-policy list");
// Unlike interactive `get`, batch must never quarantine-and-proceed —
// that would bypass a set deny policy on an unattended ingest run.
expect(fs.readFileSync(policyFile(), "utf-8")).toBe("not valid json{");
expect(
fs.readdirSync(tmpHome).find((f) => f.includes(".corrupt-")),
).toBeUndefined();
});
test("legacy allow entries migrate to read-write on batch read", () => {
fs.writeFileSync(
policyFile(),
JSON.stringify({ "github.com/foo/bar": "allow" }),
{ mode: 0o600 },
);
const r = run(["get", "--batch"], "https://github.com/foo/bar\n");
expect(r.status).toBe(0);
expect(r.stdout).toBe("read-write\n");
});
});
describe("repoPolicyTierBatch (TypeScript client)", () => {
test("maps each input url to its verdict, dedup included", () => {
expect(run(["set", "https://github.com/foo/bar", "deny"]).status).toBe(0);
expect(run(["set", "https://github.com/baz/qux", "read-only"]).status).toBe(0);
const verdicts = repoPolicyTierBatch(
[
"github.com/foo/bar", // canonical form, as memory-ingest passes it
"github.com/baz/qux",
"github.com/nope/never",
"github.com/foo/bar", // duplicate — dedup keeps ONE map entry
],
env(),
);
expect(verdicts.size).toBe(3);
expect(verdicts.get("github.com/foo/bar")).toEqual({ tier: "deny" });
expect(verdicts.get("github.com/baz/qux")).toEqual({ tier: "read-only" });
expect(verdicts.get("github.com/nope/never")).toEqual({ tier: "none" });
});
test("no store on disk: every url is tier none with no error (fast path)", () => {
const verdicts = repoPolicyTierBatch(["github.com/a/a", "github.com/b/b"], env());
expect(verdicts.get("github.com/a/a")).toEqual({ tier: "none" });
expect(verdicts.get("github.com/b/b")).toEqual({ tier: "none" });
expect(fs.existsSync(policyFile())).toBe(false);
});
test("empty url list returns an empty map without spawning", () => {
const verdicts = repoPolicyTierBatch([], env());
expect(verdicts.size).toBe(0);
});
test("corrupt store: EVERY url maps to { tier: none, error: unreadable }", () => {
fs.writeFileSync(policyFile(), "not valid json{", { mode: 0o600 });
const verdicts = repoPolicyTierBatch(["github.com/foo/bar", "github.com/baz/qux"], env());
expect(verdicts.get("github.com/foo/bar")).toEqual({ tier: "none", error: "unreadable" });
expect(verdicts.get("github.com/baz/qux")).toEqual({ tier: "none", error: "unreadable" });
});
test("store unreadable on disk (chmod 000): whole batch classified unreadable", () => {
if (!canRevokeReads()) return; // chmod is advisory here (win32, root, DAC-override containers)
expect(run(["set", "https://github.com/foo/bar", "deny"]).status).toBe(0);
fs.chmodSync(policyFile(), 0o000);
try {
const verdicts = repoPolicyTierBatch(["github.com/foo/bar"], env());
expect(verdicts.get("github.com/foo/bar")).toEqual({ tier: "none", error: "unreadable" });
} finally {
fs.chmodSync(policyFile(), 0o600);
}
});
});
// ── Normalize parity: bash normalize() ↔ lib canonicalizeRemote ─────────────
//
// bin/gstack-memory-ingest.ts produces page.git_remote via canonicalizeRemote
// (lib/gstack-memory-helpers) and then looks the policy up through
// repoPolicyTierBatch — whose bash side re-normalizes with normalize(). If
// the two functions disagree on ANY URL shape, a policy the user set via the
// script silently fails to apply to ingest (a deny that doesn't deny). The
// contract pinned here: for every shape X, `set X <tier>` followed by a batch
// lookup of canonicalizeRemote(X) returns <tier>. Bash owns normalization —
// any divergence is fixed in the SCRIPT's normalize(), never by re-normalizing
// in TypeScript.
describe("normalize parity: bash normalize() ↔ canonicalizeRemote (edge URL shapes)", () => {
// One distinct repo per shape so tiers don't overwrite each other.
const CORPUS: Array<{ shape: string; tier: "read-write" | "read-only" | "deny" }> = [
{ shape: "https://github.com/acme/plain", tier: "deny" },
{ shape: "https://github.com/acme/dotgit.git", tier: "read-only" },
{ shape: "https://github.com/acme/slash/", tier: "read-write" },
// .git + trailing slash: bash must strip the slash BEFORE the .git suffix
// (slash-first order), as canonicalizeRemote does.
{ shape: "https://github.com/acme/dotgitslash.git/", tier: "deny" },
// Uppercase .GIT: canonicalizeRemote strips case-insensitively; bash must
// lowercase before the suffix strip or the key keeps a ".git" tail.
{ shape: "https://github.com/ACME/UpperGit.GIT", tier: "read-only" },
{ shape: "git@github.com:acme/scp.git", tier: "deny" },
{ shape: "ssh://git@github.com/acme/sshurl.git", tier: "read-write" },
];
test("normalize <url> prints exactly canonicalizeRemote(url) for every corpus shape", () => {
for (const { shape } of CORPUS) {
const r = run(["normalize", shape]);
expect(r.status).toBe(0);
expect(r.stdout.trim()).toBe(canonicalizeRemote(shape));
}
});
test("a policy set via the script with shape X is found via canonicalizeRemote(X)", () => {
for (const { shape, tier } of CORPUS) {
expect(run(["set", shape, tier]).status).toBe(0);
}
const canon = CORPUS.map((c) => canonicalizeRemote(c.shape));
const verdicts = repoPolicyTierBatch(canon, env());
for (let i = 0; i < CORPUS.length; i++) {
expect(verdicts.get(canon[i])).toEqual({ tier: CORPUS[i].tier });
}
});
test("cross-shape: set through one shape, looked up through another shape of the same repo", () => {
// The store keys on the normalized form, so every spelling of the same
// repo shares one entry — set through scp form, read through https form.
expect(run(["set", "git@github.com:acme/xshape.git", "deny"]).status).toBe(0);
const canon = canonicalizeRemote("https://github.com/ACME/XShape.GIT/");
const verdicts = repoPolicyTierBatch([canon], env());
expect(verdicts.get(canon)).toEqual({ tier: "deny" });
});
});