@pmate/cli@0.45.0
CLI utilities for the pMate project
AI Security Review
scanned 18h ago · by lpm-firewall-aiThe new Electron template exposes unrestricted local process execution to a trusted renderer after login, without an approval step for each command. This presents a real agent capability risk, but I found no evidence of automatic execution or credential exfiltration.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source
Trigger
The `pmate:merry` desktop app is running, the user is logged in, and its trusted renderer invokes `local.exec`.
Impact
A compromised or misused renderer could run local commands and receive their output. The source does not show this happening automatically or sending credentials to an unrelated host.
Mechanism
The bridge accepts a caller-supplied executable, arguments, working directory, and environment, executes them locally, and returns stdout and stderr without per-command confirmation.
AI rationale
The new Electron template exposes unrestricted local process execution to a trusted renderer after login, without an approval step for each command. This presents a real agent capability risk, but I found no evidence of automatic execution or credential exfiltration.
Evidence
dist/templates/electron/main.cjsdist/templates/electron/agent-runtime.cjsdist/templates/electron/local-exec.cjs
Decision evidence
public snapshotAI called this Suspicious at 84.0% confidence as Unknown with medium false-positive risk.
Evidence for warning
- The Electron bridge is enabled only for the `pmate:merry` app and checks requests against a trusted renderer.
- After login, the bridge accepts `local.exec` requests and runs the supplied command and arguments on the local machine.
- The executor uses `execFile` without a shell, but returns command output to the caller and has no per-command confirmation.
Evidence against
- The manifest declares no install lifecycle scripts; the CLI entry point is a bin and main, and execution is guarded by `require.main === module`.
- The GitHub helper removes inherited `GH_TOKEN` and resolves a token only when a repository config names a GitHub user.
- The OpenAI client uses the configured API key and optional configured base URL; the source does not show a fixed credential redirection.
Behavioral surface
ChildProcessCryptoEnvironmentVarsEvalFilesystemNetworkShellWebSocket
Source & flagged code
11 flaggeddist/index.jsView file
90function buildSessionStorePath() {
L91: return process.env.PMATE_SESSION_FILE?.trim() || import_node_path5.default.join(import_node_os4.default.homedir(), ".pmate", "session.json");
L92: }
...
L178: const raw = (0, import_node_fs4.readFileSync)(sessionPath, "utf8");
L179: const parsed = JSON.parse(raw);
L180: const store = normalizeStore(parsed);
...
L5816: flush(true);
L5817: output.end();
L5818: } catch (error) {
...
L7692: static async toWire(msg, transport) {
L7693: if (transport === "websocket") {
L7694: const normalized2 = await _Msg.normalizeForWire(msg);
Critical
Credential Exfiltration
Source appears to send environment or credential material to an external endpoint.
dist/index.jsView on unpkg · L9090Trigger-reachable chain: scripts.start -> dist/index.js
L90: function buildSessionStorePath() {
L91: return process.env.PMATE_SESSION_FILE?.trim() || import_node_path5.default.join(import_node_os4.default.homedir(), ".pmate", "session.json");
L92: }
...
L178: const raw = (0, import_node_fs4.readFileSync)(sessionPath, "utf8");
L179: const parsed = JSON.parse(raw);
L180: const store = normalizeStore(parsed);
...
L5816: flush(true);
L5817: output.end();
L5818: } catch (error) {
...
L7692: static async toWire(msg, transport) {
L7693: if (transport === "websocket") {
L7694: const normalized2 = await _Msg.normalizeForWire(msg);
Critical
Trigger Reachable Dangerous Capability
A package entrypoint or install-time lifecycle script reaches a source file with blocking dangerous behavior.
dist/index.jsView on unpkg · L90Findings
3 Critical8 High3 Medium4 Low
CriticalCredential Exfiltrationdist/index.js
CriticalTrigger Reachable Dangerous Capabilitydist/index.js
CriticalPrevious Version Dangerous Deltadist/templates/electron/agent-runtime.cjs
HighChild Processdist/index.js
HighShelldist/index.js
HighEvaldist/index.js
HighSame File Env Network Executiondist/index.js
HighCredential Redirect Persistencedist/index.js
HighCommand Output Exfiltrationdist/index.js
HighTrigger Reachable Credential Exfiltrationdist/index.js
HighKnown Malware Source Similaritydist/index.js
MediumNetwork
Affected versions and remediation
This report applies to @pmate/cli@0.45.0.
See all versions of @pmate/cli for other recorded verdicts.
- Review the evidence and your use of @pmate/cli@0.45.0 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.