instabox@0.8.39
A cloud box for your coding agent, the instabox CLI and its agent skills in one command: npx instabox@latest
AI Security Review
scanned 7d ago · by lpm-firewall-aiThe package’s runtime behavior matches its stated CLI and agent-skill setup purpose. Authentication uses an approval-based device flow, and the reviewed telemetry is disclosed, limited to usage events, and opt-out capable.
Static reason
No blocking static signals were detected.
Trigger
The user explicitly runs the onboarding command; telemetry runs after onboarding succeeds unless disabled.
Impact
Expected account and setup actions; no credential flow to an unrelated service or harmful behavior was identified.
Mechanism
Device-code login stores Instabox credentials locally and sends them to the configured Instabox API. Onboarding can copy the package’s own skills into agent skill directories. Usage events go to PostHog.
AI rationale
The package’s runtime behavior matches its stated CLI and agent-skill setup purpose. Authentication uses an approval-based device flow, and the reviewed telemetry is disclosed, limited to usage events, and opt-out capable.
Evidence
~/.claude/skills/<skill>/...~/.agents/skills/<skill>/...~/.config/instabox/credentials.json~/.config/instabox/telemetry.json
Network endpoints2
api.instabox.shus.i.posthog.com
Decision evidence
public snapshotAI called this Clean at 91.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The manifest describes a cloud-box CLI and points its executable to bin/instabox.mjs.
- The package has no preinstall, install, or postinstall hook; its declared scripts are prepack, test, and formatting commands.
- Explicit onboarding installs this package’s own skills under ~/.claude/skills and ~/.agents/skills, and provides --no-skills to skip that setup.
- Device-code sign-in waits for user approval; saved tokens are for the Instabox server and are sent as its Bearer authorization.
- Telemetry is usage-oriented, gated to the default Instabox server, and can be disabled with INSTABOX_NO_TELEMETRY or DO_NOT_TRACK.
Evidence against
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetwork
HighEntropyStringsUrlStrings
Source & flagged code
5 flaggedbin/instabox.mjsView file
•Runtime or CLI code links a write operation to an explicit external AI-agent control path:
sync(join(dir, rel), { withfiletypes: true })
.sort((a, b) => a.name.localecompare(b.name))
.flatmap((e) => {
const r = rel ? `${rel}/${e.name}` : e.name;
return e.isdirectory() ? skillfiles(dir, r) : e.isfile() ? [r] : [];
});
}
const retired_skills = [
'instabox-migrate',
'instabox-move-all',
];
function removeretiredskills(roots) {
const removed = new set();
for (const root of roots)
for (const n of retired_skills) {
try {
High
External Ai Agent Control Surface Mutation
Runtime or CLI source writes behavior-bearing configuration into a user or project AI-agent control surface.
bin/instabox.mjsView on unpkgcli/instabox-windows-arm64.exeView file
•path = cli/instabox-windows-arm64.exe
kind = native_binary
sizeBytes = 8894976
magicHex = [redacted]
Medium
Ships Native Binary
Package ships native binary artifacts.
cli/instabox-windows-arm64.exeFindings
1 High5 Medium6 Low
HighExternal Ai Agent Control Surface Mutationbin/instabox.mjs
MediumNetwork
MediumEnvironment Vars
MediumShips Native Binarycli/instabox-windows-arm64.exe
MediumShips Build Helperskills/instabox-basic-setup/scripts/install-files.py
MediumStructural Risk Force Deep Review
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowAi Review Evidencepackage.json
LowSuspicious Lifecycle Evidencepackage.json
Affected versions and remediation
This report applies to instabox@0.8.39.
See all versions of instabox for other recorded verdicts.
- Review the evidence and your use of instabox@0.8.39 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.