registry / AI-agent capability abuse / anyray-connect / 0.11.214

anyray-connect@0.11.214

Anyray connect — points local coding tools (Claude Code, Cursor, Windsurf, SDKs) at the Anyray gateway through supported configuration and enrolled personal authentication.

AI Security Review

scanned 1mo ago · by lpm-firewall-ai

The package is not malicious under the install-hook policy because it has no automatic npm install lifecycle trigger. Its explicit-user configuration and persistent AI-agent hooks still warrant a warning due to the breadth of local control and data routing.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

A user runs the anyray-connect CLI, enrolls or supplies a gateway, and approves configuration changes.

Impact

The selected gateway can receive routed AI requests and hook-related tool output; local agent behavior remains changed until reverted.

Mechanism

AI-agent configuration, hook installation, durable runtime persistence, and gateway forwarding.

AI rationale

Evidence

dist/index.jsREADME.md~/.anyray/bin~/.claude/settings.json

Network endpoints2

localhost:8787app.anyray.ai/enroll

Decision evidence

public snapshot

AI called this Suspicious at 88.0% confidence as Dangerous Capability with low false-positive risk.

Evidence for warning

The CLI installs hooks that spawn a configured command and route local AI-tool activity through its gateway.
It persists a durable runtime and hook registrations under the user's Anyray directory.
The bundled code sends hook retrieval and optimization data to the configured gateway, including tool output.

Evidence against

package.json has no preinstall, install, or postinstall hook; prepublishOnly is not a registry-install trigger.
The behavior is reached through the package's explicit CLI entrypoint and its code refuses non-interactive changes without --yes.
No self-dependency, obfuscated payload loader, or unrelated fixed exfiltration endpoint was found.

Behavioral surface

Source

EnvironmentVarsFilesystemNetwork

Supply chain

HighEntropyStringsMinifiedPossibleObfuscationUrlStrings

ManifestNo manifest risk signals triggered.

Source & flagged code

5 flagged

dist/index.jsView file

1Source sends the broad process environment to a literal external destination. L1: #!/usr/bin/env node L2: var fL=Object.defineProperty;var M=(e,t,n)=>()=>{if(n)throw n[0];try{return e&&(t=e(e=0)),t}catch(r){throw n=[r],r}};var mL=(e,t)=>{for(var n in t)fL(e,n,{get:t[n],enumerable:!0})}... L3: `,{mode:384})}catch(c){throw await Rs(e,{recursive:!0,force:!0}).catch(()=>{}),c}return async()=>{try{JSON.parse(await xc(o,"utf-8")).nonce===s&&await Rs(e,{recursive:!0,force:!0})... L4: `,{mode:384}),await Sh(r,384);try{await Ie(r,n)}catch(o){throw await Rs(r,{force:!0}).catch(()=>{}),o}},we=async e=>{if(Kd.getStore()===!0){await vh(e);return}let t=await Rh();try{... L5: `;if(a||n.tombstoneAbsent!==!1)try{await qA(r,s)}catch(i){if(i.code!=="EEXIST")throw i}}let o=`${e}.anyray-tmp-${process.pid}-${ov

Critical

Hardcoded Runtime Data Exfiltration

Source sends credentials or rich application records to a package-controlled external receiver enabled by default.

dist/index.jsView on unpkg · L1

1#!/usr/bin/env node L2: var fL=Object.defineProperty;var M=(e,t,n)=>()=>{if(n)throw n[0];try{return e&&(t=e(e=0)),t}catch(r){throw n=[r],r}};var mL=(e,t)=>{for(var n in t)fL(e,n,{get:t[n],enumerable:!0})}... L3: `,{mode:384})}catch(c){throw await Rs(e,{recursive:!0,force:!0}).catch(()=>{}),c}return async()=>{try{JSON.parse(await xc(o,"utf-8")).nonce===s&&await Rs(e,{recursive:!0,force:!0})... L4: `,{mode:384}),await Sh(r,384);try{await Ie(r,n)}catch(o){throw await Rs(r,{force:!0}).catch(()=>{}),o}},we=async e=>{if(Kd.getStore()===!0){await vh(e);return}let t=await Rh();try{... L5: `;if(a||n.tombstoneAbsent!==!1)try{await qA(r,s)}catch(i){if(i.code!=="EEXIST")throw i}}let o=`${e}.anyray-tmp-${process.pid}-${ov()}`;try{if(await qA(o,`${JSON.stringify(t,null,2)...

High

Same File Env Network Execution

A single source file combines environment access, network access, and code or shell execution; review context before blocking.

dist/index.jsView on unpkg · L1

•Manifest-reachable source captures an API credential, sends it to a fixed unofficial gateway, and persists that redirection in agent or shell configuration. dist/index.js: var fL=Object.defineProperty;var M=(e,t,n)=>()=>{if(n)throw n[0];try{return e&&(t=e(e=0)),t}catch(r){throw n=[r],r}};var mL=(e,t)=>{for(var n in t)fL(e,n,{get:t[n],enumerable:!0})}... `,{mode:384}),await Sh(r,384);try{await Ie(r,n)}catch(o){throw await Rs(r,{force:!0}).catch(()=>{}),o}},we=async e=>{if(Kd.getStore()===!0){await vh(e);return}let t=await Rh();try{... `)},ir=(e,t={})=>{let n=e.byoUpstream,r=!n&&ne(e,"anthropic"),o=fe({...e.metadata,tool:e.metadata.tool??"claude-code"}),a=t.includeClientKey!==!1,s=t.includeAuthToken!==!1,{context... `:"";return{text:a,outcome:a===e?"unchanged":"changed"}}});import{readFile a

High

Credential Redirect Persistence

Manifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection.

dist/index.jsView on unpkg

99`;)r+=1;r-=1;continue}if(o==="/"&&e[r+1]==="*"){let a=e.indexOf("*/",r+2);r=a===-1?e.length:a+1,t+=" ";continue}t+=o}return t},NW=e=>{let t="",n=!1;for(let r=0;r<e.length;r+=1){let... L100: `,r=n.trim()?Xt(n):{};if(!lp(r))throw new Error(`${e} is not a JSON object`);return{data:r,raw:n,existed:t}},TS=(e,t,n=[])=>{if(lp(e)&&lp(t)){let r=[];for(let o of new Set([...Obje... L101: `);let a={insertSpaces:!0,tabSize:2,eol:t.raw.includes(`\r ... L109: `:` L110: ${a}${s},`,d=`${i}${l}${c}`;try{let u=Xt(d);if(!PS(u))return{ok:!1,reason:"unsupported"};for(let[p,f]of Object.entries(r))if(JSON.stringify(u[p])!==JSON.stringify(f))return{ok:!1,r... L111: `,{mode:384}),await Ie(a,o)}catch{}finally{await qS(a,{force:!0})}},kF=async(e,t)=>{await fp(cg(e,t)).catch(()=>{})},sE=async(e,t,n,r={},o={mode:"

High

Command Output Exfiltration

Source combines command execution, command-output handling, and outbound requests; review data flow before blocking.

dist/index.jsView on unpkg · L99

99Trigger-reachable command-output exfiltration chain: scripts.start -> dist/index.js L99: `;)r+=1;r-=1;continue}if(o==="/"&&e[r+1]==="*"){let a=e.indexOf("*/",r+2);r=a===-1?e.length:a+1,t+=" ";continue}t+=o}return t},NW=e=>{let t="",n=!1;for(let r=0;r<e.length;r+=1){let... L100: `,r=n.trim()?Xt(n):{};if(!lp(r))throw new Error(`${e} is not a JSON object`);return{data:r,raw:n,existed:t}},TS=(e,t,n=[])=>{if(lp(e)&&lp(t)){let r=[];for(let o of new Set([...Obje... L101: `);let a={insertSpaces:!0,tabSize:2,eol:t.raw.includes(`\r ... L109: `:` L110: ${a}${s},`,d=`${i}${l}${c}`;try{let u=Xt(d);if(!PS(u))return{ok:!1,reason:"unsupported"};for(let[p,f]of Object.entries(r))if(JSON.stringify(u[p])!==JSON.stringify(f))return{ok:!1,r... L111: `,{mode:384}),await Ie(a,o)}catch{}finally{await qS(a,{force:!

High

Trigger Reachable Command Output Exfiltration

A manifest entrypoint or package-local install chain reaches command-output exfiltration behavior.

dist/index.jsView on unpkg · L99

Source & flagged code

5 flagged · loading source

dist/index.jsView file

Critical

Hardcoded Runtime Data Exfiltration

Source sends credentials or rich application records to a package-controlled external receiver enabled by default.

dist/index.jsView on unpkg · L1

Findings

1 Critical4 High3 Medium5 Low

CriticalHardcoded Runtime Data Exfiltrationdist/index.js

HighSame File Env Network Executiondist/index.js

HighCredential Redirect Persistencedist/index.js

HighCommand Output Exfiltrationdist/index.js

HighTrigger Reachable Command Output Exfiltrationdist/index.js

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

LowNon Install Lifecycle Scripts

LowScripts Present

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to anyray-connect@0.11.214.

See all versions of anyray-connect for other recorded verdicts.

Review the evidence and your use of anyray-connect@0.11.214 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: 2026-09-13 03:13:50.833Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code