Back to skill

Security audit

TextCortex

Security checks for vulnerabilities and agentic risk

Overview

This skill needs review because it can automatically execute downloaded installation code and incorrectly treats potentially billable text generation as read-only.

Review and control CLI installation before enabling automatic setup, including its bundled-skill synchronization. Treat chat completion as external, potentially billable generation and send only content approved for OOMOL and TextCortex. The installer and hosted implementation were not verified; the artifacts do not establish malicious intent.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied command runs an Oomol installer rather than searching or reading TextCortex data. Nothing in the provided chunk establishes that this installation is a supporting dependency for TextCortex, and remote script execution is not disclosed by the description. The installer’s downstream behavior cannot be determined from this chunk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding

The supplied chunk performs login rather than the declared TextCortex data operations. Authentication could be a supporting implementation detail, but this chunk provides no evidence that the login supports TextCortex or that the declared functionality is implemented.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

The installation command pipes a remotely fetched, unpinned script directly into Bash without integrity verification or review; the PowerShell alternative has the same execution pattern. Compromise of the distribution endpoint could therefore execute attacker-controlled code with the agent process's permissions, potentially exposing local files and credentials; conditional setup reduces frequency but not the consequence.

Content

Scanner excerpt · SKILL.md (reported line 57)May include surrounding context.

The skill exposes content generation despite describing its scope as searching and reading data.

Medium
Category
Excessive Agency
Confidence
96% confidence
Finding

The manifest at L03 describes searching and reading TextCortex data, but L40 exposes create_chat_completion, which generates new content rather than retrieving existing data. The generic action runner at L30 makes this generation capability available through the skill.

Content

No source excerpt is available for this finding.

The safety documentation incorrectly classifies the untagged chat-completion action as a read.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The blanket rule that untagged actions are reads incorrectly includes create_chat_completion, which submits prompt content to a generation service and can consume paid quota. This can lead an agent to apply read-only authorization assumptions to a materially different operation, although the file does not establish destructive behavior or persistent data modification.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 78)May include surrounding context.

md
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.

- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.

## Resources