what3words

• Outcome: Review
• Latest audit: Nov 3, 2026 · 11:17 AM UTC
• Version: 1.0.4

Overview

The location lookup workflow fits its purpose, but setup executes an unverified remote installer and omits broader changes the CLI can make to agent skills.

Review and approve CLI installation separately using a verified release. Check whether it will install additional agent skills; upstream documents OO_SKILLS_SYNC_DISABLED=1 to disable startup synchronization. Submit only locations you intend to share with OOMOL and what3words. The bundled instructions were fully inspected, but external installer contents and hosted connector behavior remain unverified.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Output Handling: Unvalidated Output Injection, Cross-Context Output, Unbounded Output
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

Description-Behavior Mismatch

High

• Category: MCP Tool Poisoning
• Confidence: 98% confidence

Finding

The skill's read-oriented purpose expands into installing software when the CLI is missing, without explicitly requiring user approval. Installation executes remotely supplied code with the agent's local permissions; the missing-command prerequisite limits exposure but does not make that code trustworthy. This overlaps with the SC2 finding rather than representing a separate execution vulnerability.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High

• Category: MCP Tool Poisoning
• Confidence: 94% confidence

Finding

The supplied chunk only performs a login operation; it contains no what3words search or data-reading behavior. Authentication could support the declared purpose, but the supplied code provides no evidence that this login is related to what3words.

Content

No source excerpt is available for this finding.

External Script Fetching

High

• Category: Supply Chain
• Confidence: 99% confidence

Finding

The command pipes a mutable remote installer directly into Bash without verifying its contents, signature, or checksum; the PowerShell alternative uses the same download-and-execute pattern. If the distribution endpoint or installer supply chain is compromised, arbitrary code can run with the agent process's privileges, potentially accessing local files and credentials.

Content

Scanner excerpt · SKILL.md (reported line 59)May include surrounding context.

• oo: command not found — install the oo CLI (other platforms: https://cli.oomol.com/install-guide.md):

  bash

  curl -fsSL https://cli.oomol.com/install.sh | bash    # macOS / Linux
  

  powershell

The workflow sends location queries to external services without a location-privacy warning.

Medium

• Category: Output Handling
• Confidence: 92% confidence

Finding

The workflow instructs the agent to submit action payloads through OOMOL to what3words, potentially including precise coordinates or three-word addresses. Although the connector is identified, the skill never warns that these location queries are disclosed to external services, and its safety section labels read actions safe to run directly.

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 80)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