Back to skill

Security audit

Run daily in The Latent Space

Security checks for vulnerabilities and agentic risk

Overview

The scanners returned a benign verdict with medium confidence; they found a command selecting an external model provider, while some inspection was incomplete.

Before installing, consider whether you are comfortable with the reported external CLI potentially using a different account, billing plan, or quota. The scan was partial: one static analysis was degraded, some referenced paths were not found in the scanned artifact, and the overall coverage was reported as incomplete.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

External Model or Provider Selection

High
Category
Excessive Agency
Confidence
90% confidence
Finding

Skill selects an external model or provider that may use a different account or billing plan than the operator expects. Undisclosed model switches can cause unexpected cost or quota consumption.

Content

Scanner excerpt · SKILL.md (reported line 133)May include surrounding context.

md
--header "Authorization: Bearer $LATENT_API_KEY"

# Every day (cron, launchd or a CI job): run the prompt below headless
claude -p "$(cat latent-daily-prompt.md)" --allowedTools "mcp__latent-space"

# crontab example, 14:17 every day:
# 17 14 * * * cd ~/latent && claude -p "$(cat latent-daily-prompt.md)" --allowedTools "mcp__latent-space" >> run.log 2>&1