registry / @caeliq/llms / 1.0.71
@caeliq/llms@1.0.71
A universal LLM API transformation server
AI Security Review
scanned 26d ago · by lpm-firewall-aiThe bundles contain first-party Cursor bridge setup that modifies workspace agent configuration when bridge mode is selected. This merits a warning as an agent extension capability; the inspected source does not establish an attack.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
Agent creation in bridge mode.
Impact
Cursor agent behavior in the workspace is redirected through the package's host tools.
Mechanism
The package creates a Cursor hooks directory and writes hooks that deny built-in tools, plus workspace guidance directing use of its MCP host tools.
AI rationale
The bundles contain first-party Cursor bridge setup that modifies workspace agent configuration when bridge mode is selected. This merits a warning as an agent extension capability; the inspected source does not establish an attack.
Evidence
dist/cjs/server.cjsdist/esm/server.mjs.cursor/hooks/deny-builtin.cjs.cursor/hooks.jsonAGENTS.md
Decision evidence
public snapshotAI called this Suspicious at 86.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The package entrypoints load the inspected CJS and ESM bundles.
- Bridge mode writes Cursor hooks and an AGENTS.md file into its workspace.
- The hooks deny use of Cursor built-in tools and direct the agent to use the package's MCP host tools.
Evidence against
- The writes occur only when agent creation uses bridge mode; no npm lifecycle hook appears in the manifest.
- The inspected behavior supports the package's Cursor bridge integration; no credential exfiltration or other attack was established.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsMinifiedPossibleObfuscationUrlStrings
Source & flagged code
13 flaggeddist/esm/server.mjsView file
1var vI=Object.create;var Qh=Object.defineProperty;var xI=Object.getOwnPropertyDescriptor;var RI=Object.getOwnPropertyNames;var DI=Object.getPrototypeOf,OI=Object.prototype.hasOwnPr...
L2: -- Key to operate on
...
L58: `:case"\u2028":case"\u2029":return V(),"";case"\r":return V(),pr()===`
L59: `&&V(),"";case"1":case"2":case"3":case"4":case"5":case"6":case"7":case"8":case"9":throw Ge(V());case void 0:throw Ge(V())}return V()}function W1(){let t="",e=pr();if(!st.isHexDigit...
L60: `+s;A=S.join(E),T=`{
...
L64: `+s+A+`,
L65: `+g+"]"}return o.pop(),s=g,S}}});var Am=K((AG,Gw)=>{var J1=Hw(),V1=Ww(),Y1={parse:J1,stringify:V1};Gw.exports=Y1});var gA=K((HJ,bB)=>{bB.exports={"0":"\u2080","1":"\u2081","2":"\u2...
L66: `)||r,code:s,status:o},e.data.error)}return Object.assign({message:r,code:s,stat
Critical
Credential Exfiltration
Source appears to send environment or credential material to an external endpoint.
dist/esm/server.mjsView on unpkg · L1111${h}
L112: `+await t.crypto.sha256DigestHex(m),k=await bq(t.crypto,t.securityCredentials.secretAccessKey,i,t.region,r),g=await tc(t.crypto,k,_),y=`${tD} Credential=${t.securityCredentials.acc...
L113: To learn more about authentication and Google APIs, visit:
High
Child Process
Package source references child process execution.
dist/esm/server.mjsView on unpkg · L111Findings
3 Critical9 High4 Medium3 Low
CriticalCredential Exfiltrationdist/esm/server.mjs
CriticalTrigger Reachable Dangerous Capabilitydist/cjs/server.cjs
CriticalPrevious Version Dangerous Deltadist/cjs/server.cjs
HighChild Processdist/esm/server.mjs
HighShelldist/cjs/server.cjs
HighSame File Env Network Executiondist/esm/server.mjs
HighCommand Output Exfiltrationdist/esm/server.mjs
HighCloud Metadata Accessdist/esm/server.mjs
HighRemote Agent Bridgedist/esm/server.mjs
HighCopied Package Dependency Bridgedist/esm/server.mjs
HighTrigger Reachable Credential Exfiltrationdist/cjs/server.cjs
Affected versions and remediation
This report applies to @caeliq/llms@1.0.71.
See all versions of @caeliq/llms for other recorded verdicts.
- Review the evidence and your use of @caeliq/llms@1.0.71 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.