registry / AI-agent capability abuse / sdlc-agent-4-enterprise-server / 1.54.4

sdlc-agent-4-enterprise-server@1.54.4

Code Intelligence MCP Server — multi-agent SDLC pipeline with semantic memory, code graph, and draw.io integration

AI Security Review

scanned 2d ago · by lpm-firewall-ai

Source confirms that the explicit sandbox command tool can fall back to host execution and inherit the server environment. This warrants a warning for dangerous capability, while the inspected source does not establish a concrete malicious action; the unresolved dependency archive is not itself proof of an attack.

Static reason

One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review

Trigger

A client invokes the sandbox execution tool, and local mode is selected directly or through the configured Docker fallback.

Impact

A command can access host resources and environment values available to the server process.

Mechanism

The local executor starts the requested command through a shell and gives it the server process environment.

AI rationale

Evidence

dist/config/SandboxConfig.jsdist/modules/sandbox/ExecutionManager.jsdist/modules/sandbox/executors/LocalExecutor.jsdist/modules/sandbox/tools/tool-definitions.jsdist/modules/sandbox/tools/tool-handlers.js

Decision evidence

public snapshot

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

Evidence for warning

The package entrypoint is dist/index.js, and sandbox execution is exposed as an MCP tool that accepts a command.
Non-Linux hosts default to local execution unless full isolation is enabled; Docker unavailability can also select local mode when fallback is enabled.
The local executor inherits the server process environment and launches the command through a shell.
The sandbox handler forwards caller-supplied commands to the execution manager.
The manager creates an ephemeral session when a command is invoked without a session ID.
The unresolved receiver is an error-formatting operation that summarizes network connection attempts.
The test script logs in to a loopback service and its caller is the script’s top-level run function.

Evidence against

Command execution is an explicitly invoked sandbox tool; source inspection found no install-time execution or credential transfer to an unrelated recipient.

Behavioral surface

Source

ChildProcessCryptoDynamicRequireEnvironmentVars

Source & flagged code

19 flagged

test-index.cjsView file

Lines 1-35javascript

1const http = require('http');

2function post(url, body, headers) {

Medium

Dynamic Require

Package source references dynamic require/import behavior.

test-index.cjsView on unpkg · L1

3 return new Promise((resolve) => {

4 const u = new URL(url);

5 const data = JSON.stringify(body);

6 const req = http.request({ hostname: u.hostname, port: u.port, path: u.pathname, method: 'POST', headers: { 'Content-Type': 'application/json', 'Content-Length': Buffer.byteLength(data), ...headers } }, res => {

7 let d = ''; res.on('data', c => d += c); res.on('end', () => resolve({ status: res.statusCode, body: d }));

8 });

9 req.on('error', e => resolve({ status: 0, body: e.message }));

10 req.write(data); req.end();

11 });

12}

13async function run() {

14 // Login

15 const login = await post('http://127.0.0.1:48721/api/admin/auth/login', { username: 'admin', password: 'Admin@123456' });

Medium

Secret Pattern

Package contains a possible secret pattern.

test-index.cjsView on unpkg · L15

16 console.log('Login:', login.status);

17 if (login.status !== 200) { console.log('Login failed:', login.body); return; }

18 const token = JSON.parse(login.body).token;

19 console.log('Token:', token.substring(0, 20) + '...');

21 // Trigger full index

22 const idx = await post('http://127.0.0.1:48721/api/index/full', {}, {

23 'Authorization': 'Bearer ' + token,

24 'X-Project-Id': 'SDLC-Agents-4-Enterprise',

25 'X-Workspace-Root': 'c:\\projects\\kiro\\SDLC-Agents-4-Enterprise'

26 });

27 console.log('POST /api/index/full:', idx.status, idx.body.substring(0, 300));

29 // Wait a bit then check progress

30 await new Promise(r => setTimeout(r, 3000));

31 const prog = await new Promise((resolve) => {

32 const u = new URL('http://127.0.0.1:48721/api/index/progress');

33 http.get({ hostname: u.hostname, port: u.port, path: u.pathname, headers: { 'Authorization': 'Bearer ' + token, 'X-Project-Id': 'SDLC-Agents-4-Enterprise' } }, res => {

34 let d = ''; res.on('data', c => d += c); res.on('end', () => resolve({ status: res.statusCode, body: d }));

35 }).on('error', e => resolve({ status: 0, body: e.message }));

start.jsView file

Lines 1-24javascript

1/* This script starts the FastAPI and Next.js servers, setting up user configuration if necessary. It reads environment variables to configure API keys and other settings, ensuring that the user configuration file is created if it doesn't exist. The script also handles the starting of both servers and keeps the Node.js process alive until one of the servers exits. */

3import { join, dirname } from "path";

4import { fileURLToPath } from "url";

5import { spawn } from "child_process";

dist/modules/sandbox/executors/LocalExecutor.jsView file

Lines 37-77javascript

37 workdir: config.workdir,

38 };

39 }

40 async destroySession(_session) {

41 // Best-effort: kill any lingering child process tracked for this session.

42 return;

43 }

44 async

dist/viewer/admin/vendor/d3.min.jsView file

Lines 1-3javascript

2!function(t,n){"object"==typeof exports&&"undefined"!=typeof module?n(exports):"function"==typeof define&&

dist/engine/tools/artifact-analyzer/analyzers/FallbackAnalyzer.jsView file

Lines 11-51javascript

11 }

12 analyze(content, _options) {

13 const lines = content.split('\n');

14 const lineCount

dist/viewer/3d-force-graph.min.jsView file

1// Version 1.80.0 3d-force-graph - https://github.com/vasturiano/3d-force-graph L2: !function(e,t){"object"==typeof exports&&"undefined"!=typeof module?module.exports=t():"function"==typeof define&&define.amd?define(t):(e="undefined"!=typeof globalThis?globalThis:... L3: // <http://www.w3.org/TR/2008/REC-WCAG20-20081211/#contrast-ratiodef (WCAG Version 2) L4: // Analyze the 2 colors and returns the color contrast defined by (WCAG Version 2) L5: op.readability=function(e,t){var n=op(e),i=op(t);return(Math.max(n.getLuminance(),i.getLuminance())+.05)/(Math.min(n.getLuminance(),i.getLuminance())+.05)},op.isReadable=function(e...

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/viewer/3d-force-graph.min.jsView on unpkg · L1

dist/engine/tools/drawio-renderers.jsView file

Lines 1-21javascript

1import * as fs from 'fs';

2import { execSync } from 'child_process';

3let _cachedDrawioCliPath = null;

4export

tests/e2e/setup/global-setup.tsView file

48const serverEntry = path.resolve(__dirname, '../../../src/index.ts'); L49: serverProcess = spawn('npx', ['tsx', serverEntry], { L50: cwd: path.resolve(__dirname, '../../..'),

High

Runtime Package Install

Package source invokes a package manager install command at runtime.

tests/e2e/setup/global-setup.tsView on unpkg · L48

dist/engine/parsers/grammars/tree-sitter-go.wasmView file

•path = [redacted]-sitter-go.wasm kind = wasm_module sizeBytes = 217182 magicHex = [redacted]

Medium

Ships Wasm Module

Package ships WebAssembly modules.

dist/engine/parsers/grammars/tree-sitter-go.wasmView on unpkg

models/onnx-runtime/generate-identity-model.pyView file

•path = models/onnx-runtime/generate-identity-model.py kind = build_helper sizeBytes = 802 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

models/onnx-runtime/generate-identity-model.pyView on unpkg

servers/fastapi/tests/mocks/normalizers.pyView file

•path = [redacted].py kind = payload_in_excluded_dir sizeBytes = 489 magicHex = [redacted]

High

Payload In Excluded Dir

Package hides binary, compressed, or executable-looking payloads in test/fixture/hidden paths.

servers/fastapi/tests/mocks/normalizers.pyView on unpkg

dist/modules/memory/llm/LLMInitializer.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 12

High

Semantic Analysis Limited

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

dist/modules/memory/llm/LLMInitializer.js#virtual:normalized:round1View on unpkg

dist/engine/graph/file-resolver.jsView file

•matchType = normalized_sha256 matchedPackage = sdlc-agent-4-enterprise-server@1.53.2 matchedPath = dist/engine/graph/file-resolver.js matchedIdentity = npm:[redacted]:1.53.2 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/engine/graph/file-resolver.jsView on unpkg

dist/modules/code-intel/onboarding/WorkspaceAnalyzer.jsView file

•matchType = normalized_sha256 matchedPackage = sdlc-agent-4-enterprise-server@1.53.2 matchedPath = dist/modules/code-[redacted].js matchedIdentity = npm:[redacted]:1.53.2 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/modules/code-intel/onboarding/WorkspaceAnalyzer.jsView on unpkg

dist/modules/security/agentshield/rules/HttpEndpointRule.jsView file

•matchType = normalized_sha256 matchedPackage = sdlc-agent-4-enterprise-server@1.53.2 matchedPath = [redacted].js matchedIdentity = npm:[redacted]:1.53.2 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/modules/security/agentshield/rules/HttpEndpointRule.jsView on unpkg

check-tasks-quick.cjsView file

4patternName = generic_password severity = medium line = 4

Medium

Secret Pattern

Hardcoded password in check-tasks-quick.cjs

check-tasks-quick.cjsView on unpkg · L4

Source & flagged code

19 flagged · loading source

test-index.cjsView file

15patternName = generic_password severity = medium line = 15

Medium

Secret Pattern

Package contains a possible secret pattern.

test-index.cjsView on unpkg · L15

1const http = require('http'); L2: function post(url, body, headers) {

Medium

Dynamic Require

Package source references dynamic require/import behavior.

test-index.cjsView on unpkg · L1

start.jsView file

4import { fileURLToPath } from "url"; L5: import { spawn } from "child_process"; L6: import { randomUUID } from "crypto";

High

Child Process

Package source references child process execution.

start.jsView on unpkg · L4

Findings

12 High9 Medium8 Low

HighChild Processstart.js

HighShelldist/modules/sandbox/executors/LocalExecutor.js

HighSame File Env Network Executiondist/viewer/3d-force-graph.min.js

HighSandbox Evasion Gated Capabilitydist/engine/tools/drawio-renderers.js

HighRuntime Package Installtests/e2e/setup/global-setup.ts

HighPayload In Excluded Dirservers/fastapi/tests/mocks/normalizers.py

HighSemantic Analysis Limiteddist/modules/memory/llm/LLMInitializer.js#virtual:normalized:round1

HighKnown Malware Source Similaritydist/engine/tools/drawio-renderers.js

HighKnown Malware Source Similaritydist/modules/sandbox/executors/LocalExecutor.js

HighKnown Malware Source Similaritydist/engine/graph/file-resolver.js

Affected versions and remediation

This report applies to sdlc-agent-4-enterprise-server@1.54.4.

See all versions of sdlc-agent-4-enterprise-server for other recorded verdicts.

Review the evidence and your use of sdlc-agent-4-enterprise-server@1.54.4 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: 2026-10-21 13:39:50.785Z (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