registry / agenthub-multiagent-mcp / 1.64.0
agenthub-multiagent-mcp@1.64.0
MCP server for AgentHub multi-agent communication
AI Security Review
scanned 18d ago · by lpm-firewall-aiThe package automatically persists two Claude Code hooks in a user-wide settings file when its MCP server starts, with both features enabled by default. The reviewed activity hook stores coarse categories locally, and session capture has a separate consent guard; the remaining first-party agent setup risk warrants a warning, not a block.
Static reason
No blocking static signals were detected.; previous stored version diff introduced dangerous source
Trigger
Starting the MCP server with the default environment settings.
Impact
Claude Code sessions may continue invoking these package-owned hooks after setup, including recording coarse tool activity and running the inbox hook.
Mechanism
The server writes PostToolUse and SessionStart command hooks into ~/.claude/settings.json; environment variables can disable each hook.
AI rationale
The package automatically persists two Claude Code hooks in a user-wide settings file when its MCP server starts, with both features enabled by default. The reviewed activity hook stores coarse categories locally, and session capture has a separate consent guard; the remaining first-party agent setup risk warrants a warning, not a block.
Evidence
dist/index.jsdist/activityHookInstall.jsdist/inboxHookInstall.jsdist/activityHookCli.jsdist/activityMarker.jsdist/captureHookInstall.js~/.claude/settings.json<project>/.agenthub/activity
Decision evidence
public snapshotAI called this Suspicious at 90.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- Starting the MCP server automatically installs persistent Claude Code activity and inbox hooks in the user-wide settings file.
- Both hooks are enabled by default and can be disabled with environment variables.
- The activity hook records coarse tool categories locally; it does not record file paths, arguments, or content.
- Session capture is separately gated by server-confirmed consent and is removed when consent is absent.
Evidence against
- The reviewed activity hook records only coarse categories locally and does not itself make network requests.
- The session-capture hook follows a separate consent check and fails closed when consent cannot be confirmed.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNativeBindingsNetworkShellWebSocket
HighEntropyStringsUrlStrings
Source & flagged code
7 flaggeddist/commands/setup-shell.jsView file
25Trigger-reachable persistence chain: manifest.bin -> dist/setup.js -> dist/commands/setup-shell.js
L25: export function detectShell() {
L26: const shellPath = process.env.SHELL;
L27: if (!shellPath)
...
L35: export function rcFilePath(shell) {
L36: const home = os.homedir();
L37: return path.join(home, shell === "bash" ? ".bashrc" : ".zshrc");
L38: }
High
Trigger Reachable Persistence
A manifest entrypoint or package-local install chain reaches persistence behavior.
dist/commands/setup-shell.jsView on unpkg · L2525export function detectShell() {
L26: const shellPath = process.env.SHELL;
L27: if (!shellPath)
...
L35: export function rcFilePath(shell) {
L36: const home = os.homedir();
L37: return path.join(home, shell === "bash" ? ".bashrc" : ".zshrc");
L38: }
Medium
Install Persistence
Source writes installer persistence such as shell profile or service configuration.
dist/commands/setup-shell.jsView on unpkg · L25dist/worker.jsView file
Findings
4 High6 Medium4 Low
HighTrigger Reachable External Ai Agent Control Surface Mutationdist/worker.js
HighTrigger Reachable Persistencedist/commands/setup-shell.js
HighSemantic Analysis Limiteddist/brain/memvidBrain.js#virtual:normalized:round1
HighPrevious Version Dangerous Deltadist/setup.js
MediumNetwork
MediumEnvironment Vars
MediumInstall Persistencedist/commands/setup-shell.js
MediumShips Native Binarynative/agenthub-memvid/agenthub-memvid.win32-x64-msvc.node
MediumShips Build Helperdist/run-claude.sh
MediumStructural Risk Force Deep Review
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
Affected versions and remediation
This report applies to agenthub-multiagent-mcp@1.64.0.
See all versions of agenthub-multiagent-mcp for other recorded verdicts.
- Review the evidence and your use of agenthub-multiagent-mcp@1.64.0 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.