runlot@0.1.0-rc.81
runlot CLI — deploy apps and manage PostgreSQL
AI Security Review
scanned 1mo ago · by lpm-firewall-aiThe published CLI has no npm lifecycle hook, and the hinted Cursor MCP path is only text printed by an explicit user command. Credential use and optional crash reports stay aligned with the Runlot API and a disclosed telemetry endpoint.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
No install or import trigger writes agent configuration; mcp install runs only when the user invokes it.
Impact
No unconsented agent-control change, credential theft, or install-time execution was identified.
Mechanism
The command formats a local MCP snippet and writes it to standard output.
AI rationale
The published CLI has no npm lifecycle hook, and the hinted Cursor MCP path is only text printed by an explicit user command. Credential use and optional crash reports stay aligned with the Runlot API and a disclosed telemetry endpoint.
Decision evidence
public snapshotAI called this Clean at 86.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- package.json exposes only the runlot bin and build/test/typecheck scripts, with no preinstall, install, or postinstall hook.
- dist/index.js names Cursor and Claude config paths only as printed setup text.
- The explicit runlot mcp install command prints that snippet with console.log and returns; it does not write agent config.
Evidence against
- Crash reporting can POST scrubbed error envelopes to a Sentry ingest host unless RUNLOT_TELEMETRY or DO_NOT_TRACK disables it.
- The CLI stores its own API token in the user credentials file and talks to dash.runlot.io or a caller-set API URL.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStringsUrlStrings
NoLicense
Source & flagged code
10 flaggeddist/index.jsView file
44function gitQuery(ref, path) {
L45: const q = new URLSearchParams();
L46: if (ref !== "") q.set("ref", ref);
...
L87: async request(method, path, init) {
L88: const res = await fetch(new URL(path, this.opts.baseUrl), {
L89: ...init,
...
L95: try {
L96: body = await res.json();
L97: } catch {
...
L928: * hostname (the original request went there), and the server derives the org
L929: * and project names from it. Doing `location.replace` with the `redirect` that
L930: * comes back is the caller's job.
Critical
Credential Exfiltration
Source appears to send environment or credential material to an external endpoint.
dist/index.jsView on unpkg · L4444function gitQuery(ref, path) {
L45: const q = new URLSearchParams();
L46: if (ref !== "") q.set("ref", ref);
...
L87: async request(method, path, init) {
L88: const res = await fetch(new URL(path, this.opts.baseUrl), {
L89: ...init,
...
L95: try {
L96: body = await res.json();
L97: } catch {
...
L928: * hostname (the original request went there), and the server derives the org
L929: * and project names from it. Doing `location.replace` with the `redirect` that
L930: * comes back is the caller's job.
Critical
Remote Response Code Execution
Source passes code obtained from a remote response into a dynamic execution sink.
dist/index.jsView on unpkg · L44Findings
3 Critical5 High4 Medium6 Low
CriticalCredential Exfiltrationdist/index.js
CriticalRemote Response Code Executiondist/index.js
CriticalTrigger Reachable Dangerous Capabilitydist/index.js
HighChild Processdist/index.js
HighEvaldist/index.js
HighCredential Redirect Persistencedist/index.js
HighCommand Output Exfiltrationdist/index.js
HighTrigger Reachable Credential Exfiltrationdist/index.js
MediumDynamic Requiredist/index.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowScripts Present
Affected versions and remediation
This report applies to runlot@0.1.0-rc.81.
See all versions of runlot for other recorded verdicts.
- Review the evidence and your use of runlot@0.1.0-rc.81 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.