Summary
@anthropic-ai/claude-agent-sdk v0.2.83 was classified as CRITICAL RISK with a risk score of 6085. Sigil detected 410 findings across 40 files, covering phases including code patterns, obfuscation, install hooks, network exfiltration, credential access, provenance. Review the findings below before installing this package.
v0.2.83
25 March 2026, 06:34 UTC
by Sigil Bot
Risk Score
6085
Findings
410
Files Scanned
40
Provenance
Findings by Phase
Phase Ordering
Phases are ordered by criticality, with the most dangerous at the top. Click any phase header to expand or collapse its findings. Critical phases are expanded by default.
install-pip-setup-exec
CRITICALsetup.py executes code at install time
package/cli.js:440
See the troubleshooting guide for more information: https://aka.ms/azsdk/js/identity/azurepipelinescredential/troubleshoot`;throw oh.error(K),new Tq(K)}return oh.info("Invoking getToken() of Client Assertion Credential"),this.clientAssertionCredential.getToken(A,q)}async requestOidcToken(A,q){oh.info("Requesting OIDC token from Azure Pipelines..."),oh.info(A);let K=ZE({url:A,method:"POST",headers:jc({"Content-Type":"application/json",Authorization:`Bearer ${q}`,"X-TFS-FedAuthRedirect":"SupWhy was this flagged?
This setup.py calls subprocess, os.system, exec, or eval during package installation. Legitimate packages rarely need to execute arbitrary commands at install time. This pattern is commonly used by malicious packages to download and run payloads, exfiltrate environment variables, or establish persistence. Rated CRITICAL because it runs with the installer's full permissions.
install-pip-setup-exec
CRITICALsetup.py executes code at install time
package/cli.js:2808
`+K.join(`
`)+`
}`}return _.toString=z,_}yp1.verbose=!1});var Kcq=u((bZ$,qcq)=>{qcq.exports=Ar6;var hW_=Ym1(),RW_=Om1(),Lp1=RW_("fs");function Ar6(A,q,K){if(typeof q==="function")K=q,q={};else if(!q)q={};if(!K)return hW_(Ar6,this,A,q);if(!q.xhr&&Lp1&&Lp1.readFile)return Lp1.readFile(A,function(z,Y){return z&&typeof XMLHttpRequest<"u"?Ar6.xhr(A,q,K):z?K(z):K(null,q.binary?Y:Y.toString("utf8"))});return Ar6.xhr(A,q,K)}Ar6.xhr=function(q,K,_){var z=new XMLHttpRequest;if(z.onreadystatechange=funWhy was this flagged?
This setup.py calls subprocess, os.system, exec, or eval during package installation. Legitimate packages rarely need to execute arbitrary commands at install time. This pattern is commonly used by malicious packages to download and run payloads, exfiltrate environment variables, or establish persistence. Rated CRITICAL because it runs with the installer's full permissions.
install-pip-setup-exec
CRITICALsetup.py executes code at install time
package/cli.js:2821
`+q);let K;try{K=Vn.Metadata.fromHttp2Headers(A)}catch(Y){K=new Vn.Metadata}let _=K.getMap(),z;if(typeof _["grpc-status"]==="string"){let Y=Number(_["grpc-status"]);this.trace("received status code "+Y+" from server"),K.remove("grpc-status");let w="";if(typeof _["grpc-message"]==="string"){try{w=decodeURI(_["grpc-message"])}catch($){w=_["grpc-message"]}K.remove("grpc-message"),this.trace('received status details string "'+w+'" from server')}z={code:Y,details:w,metadata:K}}else if(this.httpStatusWhy was this flagged?
This setup.py calls subprocess, os.system, exec, or eval during package installation. Legitimate packages rarely need to execute arbitrary commands at install time. This pattern is commonly used by malicious packages to download and run payloads, exfiltrate environment variables, or establish persistence. Rated CRITICAL because it runs with the installer's full permissions.
install-makefile-curl
HIGHMakefile/script pipes remote content to shell
package/cli.js:1454
]`,`[\\[\\]\\.,\\+\\-<> \r
]`,{returnEnd:!0,relevance:0}),{className:"title",begin:"[\\[\\]]",relevance:0},{className:"string",begin:"[\\.,]",relevance:0},{begin:/(?:\+\+|--)/,contains:[q]},q]}}cOq.exports=pF9});var nOq=u((Fpw,iOq)=>{function gF9(A){if(!A)return null;if(typeof A==="string")return A;return A.source}function FF9(A){return sh1("(?=",A,")")}function dZ8(A){return sh1("(",A,")?")}function sh1(...A){return A.map((K)=>gF9(K)).join("")}function UF9(A){let q=A.COMMENT("//","$",{contains:Why was this flagged?
A script or Makefile pipes content from a remote URL directly into a shell (curl | sh or wget | bash). This is inherently dangerous because the remote content can change at any time, and the command runs with the current user's permissions. Rated HIGH because it requires manual execution (unlike install hooks) but still executes arbitrary remote code.
Badge
Markdown
[](https://sigilsec.ai/scans/EC8E636B-4EA9-42D0-81CD-CE1BDF335341)HTML
<a href="https://sigilsec.ai/scans/EC8E636B-4EA9-42D0-81CD-CE1BDF335341"><img src="https://sigilsec.ai/badge/npm/@anthropic-ai/claude-agent-sdk" alt="Sigil Scan"></a>Run This Scan Yourself
Scan your own packages
Run Sigil locally to audit any package before it touches your codebase.
Early Access
Get cloud scanning, threat intel, and CI/CD integration.
Join 150+ developers on the waitlist.
Get threat intelligence and product updates
Security research, new threat signatures, and product updates. No spam.
Other npm scans
Believe this result is incorrect? Request a review or see our Terms of Service and Methodology.