AI Security Review
scanned 11h ago · by lpm-firewall-aiReview flagged AI-agent configuration or capability changes. This remains warn-only unless evidence shows foreign-agent hijack through preinstall/install/postinstall, hidden persistence, exfiltration, remote code execution, or other concrete malicious behavior.
Static reason
One or more suspicious static signals were detected.
Trigger
User launches the MCP server and calls `setup_hermes_onboarding_agent` with an approved apply/disable request.
Impact
Can configure or roll back a package-aligned Hermes/Slack agent; no unconsented lifecycle mutation or exfiltration was confirmed.
Mechanism
Explicit MCP-controlled agent provisioning through a PATH-resolved external installer and state-registry writes.
Rationale
The package is not malicious by static source evidence, but it exposes an explicit runtime capability to provision an AI/Slack agent through an external installer. This warrants a warning rather than a publish block because no lifecycle-triggered foreign control-surface mutation or concrete malicious chain was found.
Evidence
package.jsondist/index.jsdist/server.jsdist/tools/hermesOnboardingAgent.jsdist/auth.jsdist/api.jsdist/paths.js
Network endpoints2
app.sellable.devslack.com/api/${method}
Decision evidence
public snapshotAI called this Suspicious at 88.0% confidence as Dangerous Capability with low false-positive risk.
Evidence for warning
- `dist/tools/hermesOnboardingAgent.js` exposes an MCP tool that provisions/rolls back a Hermes agent.
- The tool resolves and executes `sellable-admin-install` from `PATH` via `execFile`.
- Apply/disable require `approved: true`; the tool can update admin state/agent registry.
- `dist/server.js` activates these capabilities when the MCP server is explicitly launched.
Evidence against
- `package.json` has no `preinstall`, `install`, `postinstall`, or `prepare` hook.
- No dynamic evaluation, remote code loading, or embedded payload was found.
- The flagged secret patterns redact output and reject secret values in tool arguments.
- Network calls are authenticated product/Slack API operations, not discovered exfiltration endpoints.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsUrlStrings
Source & flagged code
2 flagged · loading sourcedist/tools/hermesOnboardingAgent.jsView file
16patternName = private_key_rsa
severity = critical
line = 16
matchedText = /-----BE...-/i,
Critical
Critical Secret
Package contains a critical-looking secret pattern.
dist/tools/hermesOnboardingAgent.jsView on unpkg · L1616patternName = private_key_rsa
severity = critical
line = 16
matchedText = /-----BE...-/i,
Critical
Secret Pattern
RSA private key in dist/tools/hermesOnboardingAgent.js
dist/tools/hermesOnboardingAgent.jsView on unpkg · L16Findings
2 Critical2 Medium4 Low
CriticalCritical Secretdist/tools/hermesOnboardingAgent.js
CriticalSecret Patterndist/tools/hermesOnboardingAgent.js
MediumNetwork
MediumEnvironment Vars
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings