Back to skill

Security audit

cargo-observability

Security checks for vulnerabilities and agentic risk

Overview

The monitoring features fit the stated purpose, but troubleshooting instructions could send sensitive commands and errors to the team without your review.

Review before installing. Require reports to be sanitized and approved before submission, removing tokens, private query values, and confidential payloads. Pin a reviewed CLI version and use narrowly scoped credentials. Preview alerts and check their targets, frequency, and recurring costs before enabling them. This review covers the supplied documentation, not the downloaded CLI or missing companion policies.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied command invokes cargo-ai workspaceManagement report create. It does not configure schedules, thresholds, monitored queries, breach actions, or alert history. Its report fields describe issue documentation rather than the declared alert-management functionality.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx @cargo-ai/cli'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

Although this is a CLI rather than an MCP server, the underlying version-pinning concern is valid: the skill recommends executing an unpinned npm package and explicitly installs @latest in its metadata. A compromised or unexpectedly changed release could execute with the invoking user's privileges and potentially access Cargo credentials; the reference to version pinning in a separate prerequisite document does not make these standalone instructions reproducible.

Content

No source excerpt is available for this finding.

Diagnostic reporting instructs sending verbatim commands and errors without a sensitive-data warning or redaction step.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The reporting instructions request exact commands and verbatim errors in a report described as readable by the team, without requiring redaction or review. This skill explicitly supports token-based login, SQL queries, and action configurations, so copied commands or errors could disclose credentials, sensitive query literals, or business data to report recipients beyond their intended audience.

Content

No source excerpt is available for this finding.