@togglhq/cli@1.13.142
Toggl CLI for Toggl 2.0 API operations.
AI Security Review
scanned 1d ago · by lpm-firewall-aiThe CLI sends runtime usage telemetry by default without an evident user-facing disclosure, which is suspicious under the review rules. The flagged GitHub-token flow is limited to explicit skill installation and uses GitHub’s API; it does not establish credential exfiltration.
Static reason
One or more suspicious static signals were detected.
Trigger
Running the CLI binary; telemetry initialization occurs before command parsing.
Impact
Toggl can observe command and action usage patterns, though the event excludes arguments and results.
Mechanism
The CLI sends a restricted tool invocation event and anonymous session identifier to Toggl’s PostHog project. An environment flag disables analytics.
AI rationale
The CLI sends runtime usage telemetry by default without an evident user-facing disclosure, which is suspicious under the review rules. The flagged GitHub-token flow is limited to explicit skill installation and uses GitHub’s API; it does not establish credential exfiltration.
Evidence
build/cli.jsbuild/program-d2cPlJ36.js
Network endpoints1
eu.i.posthog.com
Decision evidence
public snapshotAI called this Suspicious at 91.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The CLI binary initializes telemetry before parsing commands.
- Product analytics is enabled by default with a built-in Toggl PostHog key; users can disable it with TOGGL_PRODUCT_ANALYTICS=0/false/off.
- The analytics event is limited to tool name, normalized action, outcome, success, client, session ID, package version, and an error category; it excludes arguments and results.
- The CLI sends these events to the default eu.i.posthog.com endpoint.
- The GitHub-token path belongs to optional skill installation and sends the token to api.github.com.
Evidence against
- The package has no lifecycle scripts, and the exported JavaScript entry points only export createProgram.
- The analytics payload does not include command arguments, results, or credentials.
- The skills installer is invoked only through the explicit skill setup/sync commands.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
Source & flagged code
5 flaggedbuild/program-d2cPlJ36.jsView file
6import { fileURLToPath } from "node:url";
L7: import { exec, spawn } from "node:child_process";
L8: import process$1 from "node:process";
High
Child Process
Package source references child process execution.
build/program-d2cPlJ36.jsView on unpkg · L61830if (process$3.versions?.electron) parseOptions.from = "electron";
L1831: const execArgv = process$3.execArgv ?? [];
L1832: if (execArgv.includes("-e") || execArgv.includes("--eval") || execArgv.includes("-p") || execArgv.includes("--print")) parseOptions.from = "eval";
High
•Manifest-reachable source captures an API credential, sends it to a fixed unofficial gateway, and persists that redirection in agent or shell configuration.
build/program-d2cPlJ36.js:
import * as readline from "node:readline";
import { mkdir, readFile, stat, writeFile } from "node:fs/promises";
import { createInterface } from "node:readline/promises";
function readline$1() {
const cleanFn = cb(readline$1());
* Returns the width of the active readline, or 80 as default value.
function readlineWidth() {
output: readline$1().output
Findings
5 High2 Medium5 Low
HighChild Processbuild/program-d2cPlJ36.js
HighShellbuild/program-d2cPlJ36.js
HighCredential Redirect Persistencebuild/program-d2cPlJ36.js
HighCommand Output Exfiltrationbuild/program-d2cPlJ36.js
HighTrigger Reachable Command Output Exfiltrationbuild/program-d2cPlJ36.js
MediumNetwork
MediumEnvironment Vars
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to @togglhq/cli@1.13.142.
See all versions of @togglhq/cli for other recorded verdicts.
- Review the evidence and your use of @togglhq/cli@1.13.142 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.