registry / @solazah/solazah-cloud-runtime / 0.2.62
@solazah/solazah-cloud-runtime@0.2.62
Solazah 云端运行时：由 solazah-cli serve 拉起的执行端
AI Security Review
scanned 7d ago · by lpm-firewall-aiThe reviewed command execution and credential use fit the package’s stated role as a cloud runtime launched by its CLI. The install hook’s observed actions are limited to bundled platform tools and local dependency patches.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
The CLI launches the package in `serve` mode; command execution and model calls occur during runtime tasks.
Impact
No unrelated credential transfer, concealed remote execution, or harmful automatic action was found.
Mechanism
The runtime runs requested shell commands and uses account credentials with the configured Solazah API to provision model credentials. The install hook operates on bundled tools and applies local patches.
AI rationale
The reviewed command execution and credential use fit the package’s stated role as a cloud runtime launched by its CLI. The install hook’s observed actions are limited to bundled platform tools and local dependency patches.
Evidence
postinstall.mjsripgrep/<platform>/rgnode_modules dependencies targeted by local patches
Network endpoints2
api.solazah.comapi.solazah.seayona.com
Decision evidence
public snapshotAI called this Clean at 92.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The manifest describes this as a runtime launched by `solazah-cli serve`.
- The install hook checks bundled platform files, adjusts ripgrep permissions, and applies local patch files.
- Shell execution is implemented as a callable runtime method that accepts a command; it is not run by the install hook.
- The main entry dispatches to the runtime only in `serve` mode, and its parent channel requires an IPC connection.
- Account credentials are used to provision model credentials through the package’s configured Solazah API.
Evidence against
- The runtime has substantial command-execution and agent-tool capabilities, so its outputs can reach a model as part of its intended operation.
- The package defines a postinstall hook and includes native platform binaries.
Behavioral surface
ChildProcessCryptoEnvironmentVarsEvalFilesystemNetworkShellWebSocket
Source & flagged code
11 flaggedpackage.jsonView file
•scripts.postinstall = node postinstall.mjs
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg•scripts.postinstall = node postinstall.mjs
Medium
Ambiguous Install Lifecycle Script
Install-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgindex.mjsView file
11252reject(new Error("download timeout"));
L11253: }).end();
L11254: }, "doRequest");
...
L11259: runUv(args) {
L11260: return new Promise((resolve4) => {
L11261: const child = spawn(this.uvBin, args, {
L11262: env: {
...
L11267: });
L11268: let stderr = "";
L11269: child.stderr?.on("data", (c) => stderr += c.toString());
Critical
Command Output Exfiltration
Findings
2 Critical6 High5 Medium7 Low
CriticalCommand Output Exfiltrationindex.mjs
CriticalTrigger Reachable Dangerous Capabilityindex.mjs
HighInstall Time Lifecycle Scriptspackage.json
HighChild Processindex.mjs
HighShellindex.mjs
HighSame File Env Network Executionindex.mjs
HighRemote Agent Bridgeindex.mjs
HighTrigger Reachable Command Output Exfiltrationindex.mjs
MediumAmbiguous Install Lifecycle Scriptpackage.json
MediumNetwork
MediumEnvironment Vars
MediumShips Native Binaryripgrep/linux-arm64/rg
MediumStructural Risk Force Deep Review
Affected versions and remediation
This report applies to @solazah/solazah-cloud-runtime@0.2.62.
See all versions of @solazah/solazah-cloud-runtime for other recorded verdicts.
- Review the evidence and your use of @solazah/solazah-cloud-runtime@0.2.62 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.