@cursor/sdk@1.3.0
TypeScript SDK for Cursor agents.
AI Security Review
scanned 2d ago · by lpm-firewall-aiThe package’s shell execution and API-key exchange fit its Cursor agent SDK purpose, and no install hook or unrelated collector was identified. Runtime event reporting appears in the code without a disclosure or opt-out found in the package README, so this warrants a warning for telemetry rather than a publish block.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
SDK runtime events, including agent run creation, completion, and operation failures.
Impact
Runtime activity is reported to the package vendor; the reviewed code does not establish credential theft or unrelated data collection.
Mechanism
The SDK records event data and sends it to Cursor’s analytics service using the Cursor API key.
AI rationale
The package’s shell execution and API-key exchange fit its Cursor agent SDK purpose, and no install hook or unrelated collector was identified. Runtime event reporting appears in the code without a disclosure or opt-out found in the package README, so this warrants a warning for telemetry rather than a publish block.
Evidence
dist/cjs/index.js
Network endpoints1
api2.cursor.sh
Decision evidence
public snapshotAI called this Suspicious at 87.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The manifest has no preinstall, install, or postinstall hook.
- The SDK records run-created, run-completed, and operation-failure events, and sends analytics to Cursor’s default backend.
- The local shell executor is loaded through a runtime path that creates the local executor.
Evidence against
- The API key exchange sends the caller’s key to the configured backend, defaulting to Cursor’s own api2.cursor.sh service.
- Shell execution is part of the SDK’s local agent functionality; the reviewed path did not show command output being sent to an unrelated host.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStringsMinifiedPossibleObfuscationUrlStrings
NoLicense
Source & flagged code
15 flaggeddist/cjs/index.jsView file
1(()=>{"use strict";var t,r,e,n={"../../node_modules/.pnpm/@opentelemetry+api@1.9.0/node_modules/@[redacted]span_kind.js"(t,r,e){var n;e.d(r,{v:()=>n}),funct...
High
•Manifest-reachable source captures an API credential, sends it to a fixed unofficial gateway, and persists that redirection in agent or shell configuration.
dist/cjs/index.js:
(()=>{"use strict";var t,r,e,n={"../../node_modules/.pnpm/@opentelemetry+api@1.9.0/node_modules/@[redacted]span_kind.js"(t,r,e){var n;e.d(r,{v:()=>n}),funct...
High
Credential Redirect Persistence
Manifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection.
dist/cjs/index.jsView on unpkg1(()=>{"use strict";var t,r,e,n={"../../node_modules/.pnpm/@opentelemetry+api@1.9.0/node_modules/@[redacted]span_kind.js"(t,r,e){var n;e.d(r,{v:()=>n}),funct...
High
Sandbox Evasion Gated Capability
Source gates dangerous network, credential, or execution behavior behind CI, host, platform, time, or geo fingerprint checks.
Findings
3 Critical10 High5 Medium5 Low
CriticalSame File Env Network Executiondist/esm/305.js
CriticalTrigger Reachable Dangerous Capabilitydist/bundled/index.js
CriticalPrevious Version Dangerous Deltadist/cjs/305.js
HighChild Processdist/cjs/index.js
HighEvaldist/bundled/index.js
HighCredential Redirect Persistencedist/cjs/index.js
HighCommand Output Exfiltrationdist/esm/305.js
HighSandbox Evasion Gated Capabilitydist/cjs/index.js
HighObfuscated Payload Loaderdist/bundled/index.js
HighRemote Agent Bridgedist/bundled/index.js
HighCross File Remote Execution Contextdist/bundled/index.js
Affected versions and remediation
This report applies to @cursor/sdk@1.3.0.
See all versions of @cursor/sdk for other recorded verdicts.
- Review the evidence and your use of @cursor/sdk@1.3.0 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.