registry / tokenusage-cli / 0.4.0
tokenusage-cli@0.4.0
Self-hosted AI coding token usage collector for the Token Usage dashboard.
AI Security Review
scanned 4d ago · by lpm-firewall-aiThe flagged persistence and agent configuration changes are part of the documented init workflow, and init provides confirmation and integration selection. Credential use is directed to Anthropic for Claude usage, while reporting and session uploads serve the package's dashboard purpose and have documented controls.
Static reason
No blocking static signals were detected.; previous stored version diff introduced dangerous source
Trigger
Users run init to configure collection and scheduling, or explicitly run sync; package installation does not start collection.
Impact
Expected dashboard usage reporting and local integration setup; no unrelated credential transfer or hidden persistence was found.
Mechanism
Init registers an hourly sync unless --no-task is used. Sync sends usage data to the configured dashboard; Claude OAuth is used for a request to api.anthropic.com.
AI rationale
The flagged persistence and agent configuration changes are part of the documented init workflow, and init provides confirmation and integration selection. Credential use is directed to Anthropic for Claude usage, while reporting and session uploads serve the package's dashboard purpose and have documented controls.
Evidence
~/.claude/.credentials.json~/.tokenusage/~/.openclaw/openclaw.json (when the OpenClaw integration is selected)
Network endpoints2
api.anthropic.comconfigured dashboard base URL
Decision evidence
public snapshotAI called this Clean at 91.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The only lifecycle script is prepublishOnly, which runs tests.
- Interactive init asks before configuring the environment and lets users select integrations.
- The hourly schedule is registered by init, is documented, and can be skipped with --no-task.
- Run reports contain usage and run aggregates; the README documents --run-report=off.
- Claude OAuth credentials are sent as a bearer token to Anthropic's own usage endpoint.
- Session uploads exclude titles, local working directories, resume identifiers, and cache keys.
Evidence against
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsUrlStrings
Source & flagged code
7 flaggedbin/tokenusage.jsView file
5// ExperimentalWarning out of redirected sync logs.
L6: require('../src/lib/quiet-warnings').[redacted]();
L7:
Medium
Dynamic Require
Package source references dynamic require/import behavior.
bin/tokenusage.jsView on unpkg · L5src/lib/claude-categorizer.jsView file
335function defaultClaudeProjectsDir() {
L336: return path.join(os.homedir(), ".claude", "projects");
L337: }
...
L512: try {
L513: obj = JSON.parse(line);
L514: } catch (_e) {
Low
Weak Crypto
Package source references weak cryptographic algorithms.
src/lib/claude-categorizer.jsView on unpkg · L335src/lib/scheduled-task.jsView file
Findings
4 High5 Medium6 Low
HighTrigger Reachable External Ai Agent Control Surface Mutationsrc/lib/openclaw-hook.js
HighTrigger Reachable Persistencesrc/lib/scheduled-task.js
HighSemantic Analysis Limitedsrc/lib/qoder-limits.js#virtual:normalized:round1
HighPrevious Version Dangerous Deltasrc/lib/scheduled-task.js
MediumDynamic Requirebin/tokenusage.js
MediumNetwork
MediumEnvironment Vars
MediumInstall Persistencesrc/lib/scheduled-task.js
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowWeak Cryptosrc/lib/claude-categorizer.js
LowFilesystem
Affected versions and remediation
This report applies to tokenusage-cli@0.4.0.
See all versions of tokenusage-cli for other recorded verdicts.
- Review the evidence and your use of tokenusage-cli@0.4.0 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.