Security audit

cloudcc-openapi-withobject

Security checks across malware telemetry and agentic risk

Overview

This CloudCC API skill is coherent, but it ships with broad CRM-changing authority and exposed credential material that users should review before installing.

Review before installing. Rotate or remove the exposed CloudCC credentials and token, replace config.json with a template, use a least-privileged CloudCC app, restrict config and log file permissions, and require explicit user confirmation for destructive, outbound messaging, file, Chatter, and approval actions.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (8)

Context-Inappropriate Capability

Medium

Confidence: 85% confidence
Finding: The README documents a general-purpose email-sending API that goes beyond the stated object/field metadata and OpenAPI calling use case, enabling outbound communication from within the skill. In an agent context, this expands the capability surface to potential spam, phishing, data exfiltration, or unauthorized notifications if the skill is invoked with untrusted inputs or excessive permissions.

Scope Creep

Medium

Confidence: 96% confidence
Finding: The manifest declares powerful capabilities including file_read, file_write, and exec, while the listed permissions only describe network destinations. This creates a least-privilege mismatch: reviewers and users may underestimate the skill's ability to read local data, modify files, or execute commands, which materially increases the blast radius if the skill is misused or compromised.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README states that all API calls are automatically logged and describes log export/search features, but it does not warn that requests, object names, filters, IDs, and responses may contain sensitive business or personal data. In practice, such logs can become a secondary data store that leaks CRM records, identifiers, or query contents to local users, backups, or downstream tooling.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill advertises delete, update, upsert, file deletion, and approval actions, but does not provide strong user-facing warnings, confirmation requirements, or guardrails around destructive operations. In an automation setting, this can lead to accidental mass modification or deletion of business data and files through normal-looking API calls.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The documentation instructs users to store org ID, username, safety mark, client ID, and secret key in a plaintext local config file without any security warning or access-control guidance. These credentials could enable authenticated API access and compromise the tenant if the local machine, home directory, backups, or logs are exposed.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The script persists runtime-sensitive data back into config.json, including the API domain and later the access token, without any controls around file permissions or segregation of secrets. Storing bearer tokens in a general configuration file increases the chance of accidental disclosure through backups, source control mistakes, or other local processes reading the file.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The export_logs function copies the full raw log file, including API request and authentication event entries, to an arbitrary user-specified path without any warning, redaction, or permission hardening. In this skill context, logs may contain service names, object names, auth event messages, and possibly extra JSON data supplied by callers, so exporting them can unintentionally disclose operational or sensitive data to insecure locations.

Ssd 3

Medium

Confidence: 92% confidence
Finding: The skill documents automatic logging of all API calls and authentication events plus search/export features, which can capture sensitive operational metadata such as objects accessed, service names, response status, and timing. Even if tokens are not shown in the examples, these logs can expose internal business activity and become a secondary data source for attackers or unauthorized local users.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.