Back to skill

Security audit

email-thread-archaeologist

Security checks for vulnerabilities and agentic risk

Overview

This skill analyzes a chosen email file locally, but a confirmed deduplication bug can attribute repeated statements to the wrong sender.

Provide only email exports you intend the agent to read. Cross-check sender attribution, chronology, decisions, and unanswered questions against the original thread, especially repeated statements. Treat the output as a reading aid rather than an authoritative record.

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
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_read.

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.

Deduplication preserves the newest occurrence despite explicitly claiming to preserve the oldest.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

Deduplication processes strata in input order, typically newest first, so a repeated line survives under its newer attribution while its older occurrence is discarded. Sorting the retained lines afterward cannot recover the original provenance, allowing repeated or deliberately copied text to distort who first made a decision and when. This is particularly consequential for a skill whose primary purpose is reconstructing reliable decision history, although the impact is output integrity rather than code execution or data exfiltration.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.