Back to skill

Security audit

Workflow Orchestration

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local workflow-tracking library with some quality and disclosure gaps, but no evidence of hidden, destructive, or data-exfiltrating behavior.

Before installing, treat this as a lightweight local workflow-state helper rather than a full production orchestration engine. Avoid saving secrets or sensitive project context in plain JSON/YAML files, and expect a functional bug in the current exception-handling import path unless it is fixed.

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
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (6)

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The declared description presents a broader workflow orchestration engine with multi-phase coordination, routing, phase execution, and exception handling. The supplied code chunk only contains a router that selects among workflow labels such as 'hotfix', 'lightweight', and 'standard' based on task metadata. It does not execute workflows, coordinate phases, manage task progression, or provide explicit exception-handling mechanisms. Routing is consistent with part of the description, but the overall declared purpose materially overstates the implemented behavior in this code chunk.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill advertises persistence to JSON/YAML but does not warn that workflow state may be written to disk, which can cause users to store sensitive context, task metadata, or agent outputs in locations with weaker protection than expected. In orchestration contexts, persisted state may contain internal project details or secrets, so lack of disclosure increases the risk of accidental data exposure.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file includes example code that saves instance state to instance.json, which is a user-data-affecting persistence action. The guide does not include any warning or note about what data may be stored, how long it persists, or that users should avoid saving sensitive workflow contents insecurely.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The handle_exception method is documented as handling workflow exceptions, implying a usable exception-management path, but its signature references ExceptionResult without any corresponding import or definition in this file. That creates a direct code/documentation divergence because the advertised interface is not actually supported by the module as written.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

The code imports workflow instance state from a caller-provided dictionary and stores it directly in the orchestrator's internal instance registry. Although not inherently unsafe, this is a state-modifying operation that lacks any user-facing disclosure beyond a minimal docstring, so consumers are not warned that loading data will mutate active in-memory workflow state.

Content

No source excerpt is available for this finding.