Files
msd-core/sdk/src/query/check-ship-ready.ts
Tom Boucher 4e7e83bf58 fix(3587)(security): argv-based subprocess for check.ship-ready
`gsd-sdk query check.ship-ready <phase>` built a git command as a shell
string with the current branch name interpolated. Git branch names can
legally contain shell metacharacters, so a repo checked out on a
malicious branch like `foo;touch${IFS}INJ;bar` executed arbitrary shell
commands.

Vulnerability site (pre-fix):

  sdk/src/query/check-ship-ready.ts:50
    runSyncSafe(`git config --get branch.${current_branch}.merge`, cwd)
  → execSync('git config --get branch.foo;touch${IFS}INJ;bar.merge')
  → /bin/sh -c parses three commands; the middle one runs `touch INJ`
    in the project dir and creates the sentinel file.

Manually reproduced on git 2.53.0:
  - refname `foo;touch${IFS}INJ;bar` is accepted by `git check-ref-format`
    and by `git checkout -b`.
  - `current_branch` returned from `git rev-parse --abbrev-ref HEAD`
    contains the metacharacters verbatim.
  - Interpolation into the buggy execSync call creates the sentinel.

Fix:

- Replace `runSyncSafe(cmd: string, cwd)` (execSync, shell-string) with
  `runArgvSafe(file, args: readonly string[], cwd)` (execFileSync,
  argv-based, no shell).
- Same shape for the boolean wrapper: `boolArgvSafe`.
- Convert all 7 subprocess sites in the module to argv form:
  - `git status --porcelain`
  - `git rev-parse --abbrev-ref HEAD`
  - `git config --get branch.<name>.merge`   ← the interpolation site
  - `git rev-parse --verify main`
  - `git remote`
  - `gh --version`
  - `which gh`
- Shell is never invoked. Branch names — even ones with `;`, `$IFS`,
  backticks, `$()` — are passed as a single argv element and treated
  as opaque data.

Regression test (`sdk/src/query/check-ship-ready.test.ts`):

- `#3587: branch name with shell-injection payload does not execute
  injected command` — creates a real git repo, checks out the proven
  exploit branch `foo;touch${IFS}INJECTED_BY_3587;bar`, runs
  checkShipReady, and asserts the sentinel file does NOT exist. This
  test FAILS on the unfixed code (verified pre-implementation) and
  PASSES on the fixed code — true red→green TDD.
- `#3587: round-trips a metacharacter branch name verbatim in
  current_branch` — positive proof the branch name survives argv as
  data (would fail if a future change re-introduces shell quoting).
- `#3587: gh probe does not invoke a shell` — locks the gh path
  against a future regression that might add an interpolation site.

Validation:
- SDK unit suite via vitest: 1,869/1,869 pass.
- Full root suite via gsd-test-both (per CLAUDE.md): 10,676/10,676
  on Mac AND 10,676/10,676 on Linux Docker, zero cross-platform diff.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-15 20:42:52 -04:00

130 lines
4.6 KiB
TypeScript

/**
* Ship preflight checks (`check.ship-ready`).
*
* Consolidates git/gh checks from `ship.md` into a single structured query.
* All subprocess calls are wrapped in try/catch — never throws on git/gh failures.
* See `.planning/research/decision-routing-audit.md` §3.9.
*
* #3587: every subprocess call uses argv-based execFileSync — never a
* shell-string execSync. Git branch names are repository-controlled data
* and can legally contain metacharacters (`;`, `$`, backticks, etc.); a
* shell-string `git config --get branch.${current_branch}.merge` allowed
* arbitrary command injection from a malicious branch name. Passing args
* as argv elements means the shell is never invoked.
*/
import { execFileSync } from 'node:child_process';
import { GSDError, ErrorClassification } from '../errors.js';
import { normalizePhaseName } from './helpers.js';
import { checkVerificationStatus } from './check-verification-status.js';
import type { QueryHandler } from './utils.js';
/**
* Run a subprocess via argv (NEVER a shell string). Returns trimmed stdout
* on success or null on any failure (non-zero exit, missing binary, etc.).
* The pre-#3587 helper used `execSync(cmd, …)` which spawned `/bin/sh -c`
* and parsed `cmd` as shell syntax — that path is gone.
*/
function runArgvSafe(file: string, args: readonly string[], cwd: string): string | null {
try {
return execFileSync(file, args, {
cwd,
encoding: 'utf-8',
stdio: ['pipe', 'pipe', 'pipe'],
}).trim();
} catch {
return null;
}
}
function boolArgvSafe(file: string, args: readonly string[], cwd: string): boolean {
return runArgvSafe(file, args, cwd) !== null;
}
export const checkShipReady: QueryHandler = async (args, projectDir) => {
const raw = args[0];
if (!raw) {
throw new GSDError('phase number required for check ship-ready', ErrorClassification.Validation);
}
normalizePhaseName(raw); // validate format
const blockers: string[] = [];
// git checks — all wrapped in try/catch via helpers, all argv-based.
const porcelain = runArgvSafe('git', ['status', '--porcelain'], projectDir);
const clean_tree = porcelain !== null && porcelain === '';
const current_branch = runArgvSafe('git', ['rev-parse', '--abbrev-ref', 'HEAD'], projectDir);
const on_feature_branch =
current_branch !== null &&
current_branch !== 'main' &&
current_branch !== 'master';
// Determine base branch
let base_branch: string | null = null;
if (current_branch) {
// #3587: branch name passed as a single argv element — git treats it
// as data, the shell is never invoked, no interpolation possible.
const mergeRef = runArgvSafe(
'git',
['config', '--get', `branch.${current_branch}.merge`],
projectDir,
);
if (mergeRef) {
base_branch = mergeRef.replace('refs/heads/', '');
} else {
// Fallback: check if 'main' branch exists, else 'master'
const mainExists = boolArgvSafe('git', ['rev-parse', '--verify', 'main'], projectDir);
base_branch = mainExists ? 'main' : 'master';
}
}
const remoteOut = runArgvSafe('git', ['remote'], projectDir);
const remote_configured = remoteOut !== null && remoteOut.trim().length > 0;
// gh availability — argv as well so a future change that interpolates
// a user-controlled value into the probe cannot silently introduce a
// new injection seam.
const gh_available =
boolArgvSafe('gh', ['--version'], projectDir) ||
boolArgvSafe('which', ['gh'], projectDir);
// gh_authenticated: advisory — skip actual auth check to avoid slow network call
const gh_authenticated = false;
// Verification status
let verification_passed = false;
try {
const verRes = await checkVerificationStatus([raw], projectDir);
const vdata = verRes.data as Record<string, unknown>;
const status = String(vdata.status ?? '').toLowerCase();
verification_passed = status === 'pass' || status === 'passed';
} catch {
verification_passed = false;
}
// Collect blockers
if (!verification_passed) blockers.push('verification status is not passed');
if (!clean_tree) blockers.push('working tree is not clean (uncommitted changes)');
if (!on_feature_branch) blockers.push('not on a feature branch (currently on main/master or unknown)');
if (!remote_configured) blockers.push('no git remote configured');
const ready = verification_passed && clean_tree && on_feature_branch && remote_configured;
return {
data: {
ready,
verification_passed,
clean_tree,
on_feature_branch,
current_branch,
base_branch,
remote_configured,
gh_available,
gh_authenticated,
blockers,
},
};
};