@ory/claude-agent-sdk@1.8.0
Ory Agent Security for the Anthropic Claude Agent SDK — per-tool authorization, activity auditing, and identity propagation via PreToolUse hooks and canUseTool. Built on @ory/argus.
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe package enables dependency postinstall analytics by default and configures an endpoint; the event includes the installer’s Git email domain. This is install-time telemetry with an identifying organizational detail, so it warrants a warning.
Static reason
No blocking static signals were detected.
Trigger
Dependency postinstall runs during installation when analytics are enabled and an endpoint is configured.
Impact
The configured analytics service receives install telemetry that includes an organizational identifier without a separate installer prompt.
Mechanism
The reova tracker reads the email domain from ~/.gitconfig, adds it to an npm-install event, and POSTs the event to the configured telemetry endpoint.
AI rationale
The package enables dependency postinstall analytics by default and configures an endpoint; the event includes the installer’s Git email domain. This is install-time telemetry with an identifying organizational detail, so it warrants a warning.
Evidence
package.json.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/index.js.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/tracker.js.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/collector.js.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/sender.js~/.gitconfig
Network endpoints1
telemetry.reo.dev/data
Decision evidence
public snapshotAI called this Suspicious at 92.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The manifest enables reova analytics and configures https://telemetry.reo.dev/data.
- reova's postinstall runs its tracker automatically during installation.
- The tracker collects an install event and sends it when the package has analytics enabled and an endpoint configured.
- The event includes the domain parsed from the installer's Git email in ~/.gitconfig.
- The sender POSTs the collected event to the configured endpoint.
Evidence against
- The package has no own install lifecycle script; the automatic behavior comes from its reova dependency.
- The collector sends the email domain rather than the full email address and excludes environment variables and file paths.
- The endpoint is explicitly declared in the package manifest, and the tracker supports REOVA_ANALYTICS=false and DO_NOT_TRACK opt-outs.
Behavioral surface
Source & flagged code
6 flaggedpackage.jsonView file
11"reova": {
"enabled": true,
"endpoint": "https://telemetry.reo.dev/data"
},
"main": "dist/index.js",
Medium
Ai Review Evidence
The manifest enables reova analytics and configures https://telemetry.reo.dev/data.
package.jsonView on unpkg · L11.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/index.jsView file
17const trackingPromise = delay(START_DELAY_MS)
.then(() => new Tracker().track())
.catch(() => {
// Never surface errors to the install.
});
Promise.race([track
Medium
Suspicious Lifecycle Evidence
reova's postinstall runs its tracker automatically during installation.
.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/index.jsView on unpkg · L17Findings
6 Medium1 Low
MediumAi Review Evidencepackage.json
MediumSuspicious Lifecycle Evidence.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/index.js
MediumAi Review Evidence.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/tracker.js
MediumAi Review Evidence.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/collector.js
MediumAi Review Evidence.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/collector.js
MediumAi Review Evidence.lpm-dependency-evidence-a2a51aa2-dc23-4a74-8dc0-5c8442d95ddc/1/npm-linux/src/sender.js
LowScripts Present
Affected versions and remediation
This report applies to @ory/claude-agent-sdk@1.8.0.
See all versions of @ory/claude-agent-sdk for other recorded verdicts.
- Review the evidence and your use of @ory/claude-agent-sdk@1.8.0 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.