Back to skill

Security audit

Proactive Ops Monitor

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent workspace operations monitor, but it can surface recent agent-memory content and persist alert/config files in the workspace.

Install this only in workspaces where it is acceptable for an ops tool to read recent agent memory and show selected draft, decision, or correction content in dashboards or suggestions. Treat alert_config as a persistent settings change, and review or remove workspace alert/config files if they contain operational details you do not want retained.

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
Findings (1)

other

Warning
Location
scripts/run.py:86
Finding

Agent Memory Contents Exposed Through Suggestion Output

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill behavior clearly includes reading and writing workspace files such as memory/wal.jsonl, notes/areas/open-loops.md, memory/ops-alerts.jsonl, and proactive-ops-config.json, but it does not declare an overall tool/file-access scope beyond per-tool entries. That mismatch weakens least-privilege review and can cause users or hosting platforms to underestimate the skill’s access to sensitive workspace state and its ability to persist data.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill is presented as a monitoring/dashboard utility, but the alert_config action allows persistent modification of proactive-ops-config.json. In an agent environment, that expands the tool from read-only observability into state-changing behavior, which can be abused to disable alerts, alter thresholds, or suppress operational signals without clear user awareness.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

Writing to a persistent configuration file is not clearly necessary for a skill whose stated purpose is operational monitoring. This mismatch increases risk because consumers may grant or invoke the skill expecting passive telemetry, while it can actually alter future behavior and reduce safety visibility by changing alert or suggestion settings.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

CONFIG_FILE.write_text(...) overwrites persistent configuration immediately, with no confirmation, backup, concurrency protection, or warning to the user. An unintended or malicious invocation could silently replace thresholds or settings, causing alert suppression, noisy misconfiguration, or operational blind spots that persist across runs.

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 monitoring and alerting benefits, but the user-facing description does not prominently warn that alerting persists monitoring data into workspace files. This can expose conversation-derived operational metadata, open-loop/task details, and usage patterns in files that may later be read by other skills, users, or automation, increasing privacy and cross-skill data exposure risk.

Content

No source excerpt is available for this finding.