registry / knight-agent / 1.3.4

knight-agent@1.3.4

Knight-agent智能体架构

AI Security Review

scanned 1d ago · by lpm-firewall-ai

No confirmed unconsented attack surface is established. The bundle performs configured LLM HTTP requests only when the caller configures an agent and sends a prompt.

Static reason

One or more suspicious static signals were detected.

Trigger

Caller invokes `Agent.OnSetConfig` and `Agent.OnSend`.

Impact

Expected transfer of caller-provided prompts and API key to the selected LLM endpoint.

Mechanism

Configured LLM API requests via fetch.

Rationale

The suspicious signals resolve to obfuscation and an explicitly invoked registry-token script, not install-time or covert runtime behavior. Source inspection supports a user-configured LLM client with no confirmed exfiltration, persistence, or execution chain.

Evidence

package.jsonindex.jsreadme.mdutility/NetworkUtility.d.tscomponent/Agent.d.tscomponent/Tool.d.ts

Network endpoints2

api.openai.com/v1/chat/completionsdashscope.aliyuncs.com/compatible-mode/v1/chat/completions

Decision evidence

public snapshot

AI called this Clean at 91.0% confidence as Benign with medium false-positive risk.

Evidence for block

`index.js` is heavily obfuscated.
`package.json` contains an explicit `authToken` script that sets an npm registry credential.

Evidence against

`package.json` has no preinstall, install, postinstall, or prepare hook.
`index.js` has no direct child-process, filesystem-write, eval, dynamic import, or require primitive.
`index.js` uses fetch for the documented LLM client flow only after caller configuration and send actions.
`readme.md` and `.d.ts` define caller-supplied LLM keys, endpoints, prompts, and tool handlers; no credential harvesting is exposed.
No AI-agent config paths, persistence paths, destructive commands, or hidden payload loaders were found.

Behavioral surface

SourceNo risky source behavior triggered.

Supply chain

HighEntropyStringsMinifiedTrivial

ManifestNo manifest risk signals triggered.

scanned 1 file(s), 77.0 KB of source

Source & flagged code

2 flagged

package.jsonView file

Lines 1-19json

2 "name": "knight-agent",

3 "version": "1.3.4",

4 "description": "Knight-agent智能体架构",

5 "main": "index.js",

6 "scripts": {

7 "test": "echo \"Error: no test specified\" && exit 1",

8 "authToken": "npm config set //registry.npmjs.org/:_authToken=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",

Critical

Critical Secret

Package contains a critical-looking secret pattern.

package.jsonView on unpkg · L8

Critical

Secret Pattern

npm access token in package.json

package.jsonView on unpkg · L8

9 "login": "npm login",

10 "publish": "npm publish"

11 },

12 "keywords": [

13 "knight",

14 "agent",

15 "knight-agent"

16 ],

17 "author": "ZiXuWang",

18 "license": "MIT"

19}

Source & flagged code

2 flagged · loading source

package.jsonView file

8patternName = npm_token severity = critical line = 8 matchedText = "authTok...U2",

Critical

Critical Secret

Package contains a critical-looking secret pattern.

package.jsonView on unpkg · L8

8patternName = npm_token severity = critical line = 8 matchedText = "authTok...U2",

Critical

Secret Pattern

npm access token in package.json

package.jsonView on unpkg · L8

Findings

2 Critical2 Low

CriticalCritical Secretpackage.json

CriticalSecret Patternpackage.json

LowScripts Present

LowHigh Entropy Strings