Back to skill

Security audit

Deepgram

Security checks for vulnerabilities and agentic risk

Overview

The Deepgram workflow fits its purpose, but setup can execute unverified software with broader effects on installed agent skills than this skill explains.

Review and install the CLI separately before using this skill. Consider disabling automatic skill synchronization using the publisher-documented OO_SKILLS_SYNC_DISABLED setting in its [command reference](https://github.com/oomol-lab/oo-cli/blob/main/docs/commands.md). Connect only the intended Deepgram account through OOMOL. The entire bundled artifact was inspected, but the remote installers and hosted connector implementation were not verified.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk performs remote script execution rather than searching or reading Deepgram data. Nothing in the provided content establishes this installation as a supporting requirement for the declared workflow. The downloaded script's contents and subsequent effects are not shown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Authentication could support a Deepgram integration, but the supplied chunk consists solely of a login command and establishes no connection to Deepgram. As provided, its primary behavior differs materially from the declared purpose. The command alone does not establish credential theft or other malicious behavior.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

The curl-to-bash pipeline immediately executes mutable remote content without pinning or verifying its integrity. A compromised installer host or publishing pipeline could therefore run arbitrary commands with the invoking user's privileges; HTTPS protects transport but not against a compromised publisher.

Content

Scanner excerpt · SKILL.md (reported line 62)May include surrounding context.

The setup workflow executes downloaded installer scripts without warning about running third-party code or modifying the local system.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

When the CLI is missing, the skill directs execution of downloaded shell or PowerShell code without an explicit installation approval step or an integrity check. This can modify the local system with the executing user's privileges; the conditional setup context reduces exposure but does not make the downloaded code trustworthy.

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 83)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