registry / AI-agent capability abuse / clavue-v1 / 1.10.3

clavue-v1@1.10.3

Clavue v1: lightweight four-slot (main/light/plan/review) multi-model coding CLI on the execution-first engine — overlapped cross-family review, official Clavue cloud preset (api.clavue.com) alongside any Messages-compatible provider, learned context ceil

AI Security Review

scanned 1mo ago · by lpm-firewall-ai

The critical scanner signal is not supported as credential theft: inspected requests target the user-selected provider URL. The explicit provider setup can alter external agent-related configuration, so it warrants a warning rather than a block.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

A user runs the Clavue provider setup or activates a provider profile.

Impact

It can redirect compatible coding tools to the selected provider and credential.

Mechanism

Provider-profile persistence modifies local routing configuration.

AI rationale

Evidence

dist/provider-setup.js.clavue/settings.json.clavue/.credentials.json.ccjk/config.toml.ufomiao/zcf/config.toml

Decision evidence

public snapshot

AI called this Suspicious at 84.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

The provider profile persistence flow updates compatible external tool configuration, not just Clavue-owned files.
The package bundles a powerful coding-agent runtime and can set provider credentials and model-routing environment values.

Evidence against

package.json defines no preinstall, install, or postinstall hook.
The configuration flow is reached through the package CLI and interactive provider setup, not during npm installation.
The reviewed network validation requests use the selected provider profile URL; no concrete secret exfiltration endpoint was established.

Behavioral surface

Source

ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell

Supply chain

HighEntropyStringsMinifiedPossibleObfuscationUrlStrings

Manifest

NoLicense

Source & flagged code

11 flagged

dist/mao-command.jsView file

4var __create=Object.create;var __defProp=Object.defineProperty;var __getOwnPropDesc=Object.getOwnPropertyDescriptor;var __getOwnPropNames=Object.getOwnPropertyNames;var __getProtoO... L5: GFS4: `),console.error(m)},"debug"));fs6[gracefulQueue]||(queue=global[gracefulQueue]||[],publishQueue(fs6,queue),fs6.close=(function(fs$close){function close(fd,cb){return fs$clos... L6: `,"utf8"),process4.platform!=="win32")try{fs.chmodSync(filePath,384)}catch{}}__name(defaultWriteJsonFile,"defaultWriteJsonFile");function defaultSafeReadFile(filePath){try{return f...

High

Child Process

Package source references child process execution.

dist/mao-command.jsView on unpkg · L4

node_modules/@anthropic-ai/sandbox-runtime/dist/cli.jsView file

44.option('-s, --settings <path>', 'path to config file (default: ~/.srt-settings.json)') L45: .option('-c <command>', 'run command string directly (like sh -c), no escaping applied') L46: .option('--control-fd <fd>', 'read config updates from file descriptor (JSON lines protocol)', parseInt)

High

Shell

Package source references shell execution.

node_modules/@anthropic-ai/sandbox-runtime/dist/cli.jsView on unpkg · L44

dist/provider-router.jsView file

3// clavue v1.10.3 provider-router (built from source) L4: var __create=Object.create;var __defProp=Object.defineProperty;var __getOwnPropDesc=Object.getOwnPropertyDescriptor;var __getOwnPropNames=Object.getOwnPropertyNames;var __getProtoO... L5: `?(templateIndex-=1,rawIndex+=1):nextRawCharacter==="u"&&rawTemplate[rawIndex+2]==="{"?rawIndex=rawTemplate.indexOf("}",rawIndex+3):rawIndex+=ESCAPE_LENGTH[nextRawCharacter]??1}}le... L6: `]),ESCAPE_LENGTH={x:3,u:5},concatTokens=__name((tokens,nextTokens,isSeparated)=>isSeparated||tokens.length===0||nextTokens.length===0?[...tokens,...nextTokens]:[...tokens.slice(0,... L7: It must be "${optionName}.stdout", "${optionName}.stderr", "${optionName}.all", "${optionName}.ipc", or "${optionName}.fd3", "${optionName}.fd4" (and so on).`);if(fdNumber>=o

Critical

Credential Exfiltration

Source appears to send environment or credential material to an external endpoint.

dist/provider-router.jsView on unpkg · L3

32${originalMessage}`}return cwdStat.isDirectory()?originalMessage:`The "cwd" option is not a directory: ${cwd2}. L33: ${originalMessage}`},"fixCwdError")}});import path5 from"node:path";import process6 from"node:process";var import_cross_spawn,normalizeOptions,addDefaultOptions,getEnv,init_options... L34: `,LF_BINARY=LF.codePointAt(0),CR="\r",CR_BINARY=CR.codePointAt(0)}});function isStream(stream4,{checkOpen=!0}={}){return stream4!==null&&typeof stream4=="object"&&(stream4.writable... ... L45: Instead, \`yield\` should either be called with a value, or not be called at all. For example: L46: if (condition) { yield value; }`)},"validateEmptyReturn")}});import{Buffer as Buffer3}from"node:buffer";import{StringDecoder as StringDecoder2}from"node:string_decoder";var getEnco... L47: `):[]}}

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/provider-router.jsView on unpkg · L32

dist/vendor/ripgrep/x64-darwin/rgView file

•path = dist/vendor/ripgrep/x64-darwin/rg kind = native_binary sizeBytes = 4442900 magicHex = [redacted]

Medium

Ships Native Binary

Package ships native binary artifacts.

dist/vendor/ripgrep/x64-darwin/rgView on unpkg

install.shView file

•path = install.sh kind = build_helper sizeBytes = 1668 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

install.shView on unpkg

dist/cli.jsView file

•path = dist/cli.js kind = oversized_source_file sizeBytes = 14597774 magicHex = [redacted]

Medium

Oversized Source File

Package contains source files above the normal full-analysis size ceiling.

dist/cli.jsView on unpkg

•path = dist/cli.js kind = oversized_cli_entrypoint sizeBytes = 14597774 magicHex = [redacted]

Medium

Oversized Cli Entrypoint

Package contains an oversized executable-looking CLI entrypoint.

dist/cli.jsView on unpkg

node_modules/@anthropic-ai/sandbox-runtime/dist/sandbox/http-proxy.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 16

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

node_modules/@anthropic-ai/sandbox-runtime/dist/sandbox/http-proxy.js#virtual:normalized:round1View on unpkg

Source & flagged code

11 flagged · loading source

dist/mao-command.jsView file

High

Child Process

Package source references child process execution.

dist/mao-command.jsView on unpkg · L4

node_modules/@anthropic-ai/sandbox-runtime/dist/cli.jsView file

High

Shell

Package source references shell execution.

node_modules/@anthropic-ai/sandbox-runtime/dist/cli.jsView on unpkg · L44

Findings

1 Critical6 High7 Medium5 Low

CriticalCredential Exfiltrationdist/provider-router.js

HighChild Processdist/mao-command.js

HighShellnode_modules/@anthropic-ai/sandbox-runtime/dist/cli.js

HighSame File Env Network Executiondist/provider-router.js

HighCommand Output Exfiltrationdist/provider-router.js

HighCloud Metadata Accessdist/provider-router.js

HighSemantic Analysis Limitednode_modules/@anthropic-ai/sandbox-runtime/dist/sandbox/http-proxy.js#virtual:normalized:round1

MediumNetwork

MediumEnvironment Vars

MediumShips Native Binarydist/vendor/ripgrep/x64-darwin/rg

MediumShips Build Helperinstall.sh

Medium

Affected versions and remediation

This report applies to clavue-v1@1.10.3.

See all versions of clavue-v1 for other recorded verdicts.

Review the evidence and your use of clavue-v1@1.10.3 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 02:14:43.421Z (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