@agentvault/agentvault@0.23.65
End-to-end encrypted messaging for AI agents: connects an agent to its owner, rooms and other agents on AgentVault.
AI Security Review
scanned 2h ago · by lpm-firewall-aiThe plugin automatically reports operational telemetry, including model and token usage and HTTP URLs, to its configured AgentVault service. The README discloses the telemetry and its purpose, but no opt-out control was found, so this warrants a warning rather than a block.
Static reason
One or more suspicious static signals were detected.
Trigger
An enrolled AgentVault channel connects with persisted device credentials; message and agent operations then generate spans.
Impact
The AgentVault service receives ongoing operational details about the connected agent. This is a privacy risk, though the telemetry is documented and no prompt or message content collection was identified.
Mechanism
The plugin batches telemetry spans and periodically POSTs them with the device JWT to the configured API endpoint. Fields include model and token usage, tool activity, and HTTP URLs.
AI rationale
The plugin automatically reports operational telemetry, including model and token usage and HTTP URLs, to its configured AgentVault service. The README discloses the telemetry and its purpose, but no opt-out control was found, so this warrants a warning rather than a block.
Evidence
dist/index.jsREADME.md
Network endpoints1
api.agentvault.chat/api/v1/telemetry/ingest (default; host follows configured API URL)
Decision evidence
public snapshotAI called this Suspicious at 87.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The README says the plugin automatically instruments message operations with telemetry for trust scoring and the observability dashboard.
- When the enrolled channel has a device JWT and hub ID, it creates a telemetry reporter using the configured API URL and device JWT.
- The reporter starts automatic flushing every 30 seconds and POSTs spans to the telemetry ingest endpoint.
- LLM spans include model names and input/output token counts; HTTP spans include request URLs.
Evidence against
- The README discloses the telemetry and describes its product purpose.
- Telemetry goes to the configured AgentVault API, whose default is https://api.agentvault.chat.
- The inspected telemetry fields contain operational metadata; no prompt or message content collection was identified.
- The manifest’s prepublishOnly script only runs the package build.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetwork
Source & flagged code
9 flaggeddist/cli.jsView file
93913});
L93914: import { execSync } from "node:child_process";
L93915: import { existsSync as existsSync2, readFileSync as readFileSync4, writeFileSync, chmodSync, readdirSync, statSync } from "node:fs";
High
94026try {
L94027: execSync("curl -sf --max-time 3 http://127.0.0.1:18789/", { stdio: "pipe" });
L94028: return { name: "Gateway port (18789)", status: "pass", message: "Responding" };
...
L94040: }
L94041: const home = process.env.HOME ?? "";
L94042: const warnings = [];
High
Same File Env Network Execution
A single source file combines environment access, network access, and code or shell execution; review context before blocking.
dist/cli.jsView on unpkg · L94026120}
L121: function isPrivatePairHumanMessage(senderIsAgent, roomAudience) {
L122: return senderIsAgent === false && roomAudience === "private_pair";
...
L132: // ../../node_modules/libsodium-sumo/dist/modules-sumo-esm/libsodium-sumo.mjs
L133: var __filename, __dirname, url, path, Module, Module, root, window_, crypto_, randomValuesStandard, crypto2, randomValueNodeJS, _Module, libsodium_sumo_default;
L134: var init_libsodium_sumo = __esm({
...
L227: Module2.useBackupModule = function() {
L228: return new Promise(function(resolve6, reject2) {
L229: var Module3 = {};
...
L280: quit_2 = (status, toThrow) => {
L281: process.exitCode = status;
L282: throw toThrow;
Findings
7 High6 Medium5 Low
HighChild Processdist/cli.js
HighShell
HighSame File Env Network Executiondist/cli.js
HighCredential Redirect Persistencedist/index.js
HighCommand Output Exfiltrationdist/cli.js
HighCross File Remote Execution Contextdist/cli.js
HighSemantic Analysis Limiteddist/cli.js
MediumDynamic Requiredist/openclaw-entry.js
MediumNetwork
MediumEnvironment Vars
MediumOversized Source Filedist/index.js
MediumOversized Cli Entrypointdist/cli.js
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
Affected versions and remediation
This report applies to @agentvault/agentvault@0.23.65.
See all versions of @agentvault/agentvault for other recorded verdicts.
- Review the evidence and your use of @agentvault/agentvault@0.23.65 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.