T08 · Insecure Dependencies
- Location
SKILL.md:40- Finding
Unpinned Third-Party Package Is Automatically Downloaded and Executed
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
The skill’s reader-library purpose is coherent, but its setup asks users to run an unpinned remote MCP package and pass a reusable token in plaintext command arguments.
Review this before installing. Prefer a pinned and reviewed `sombra-mcp` version, avoid placing long-lived tokens directly in shared config or process arguments, use OAuth or a secret manager if available, issue a narrow-scope token, and run the MCP server with only the local and network access it needs.
SKILL.md:40Unpinned Third-Party Package Is Automatically Downloaded and Executed
SKILL.md:36Personal Access Token Is Stored in Configuration and Passed Through Process Arguments
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
## Setup
### 1. Get a Personal Access Token
Sign up at [sombra.so](https://sombra.so) and create a token in **Settings > Access Tokens**.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
### 1. Get a Personal Access Token
Sign up at [sombra.so](https://sombra.so) and create a token in **Settings > Access Tokens**.
### 2. Configure OpenClaw
The skill instructs users to launch an MCP server via npx without pinning an exact package version or integrity mechanism. This creates a supply-chain risk: a future malicious or compromised package release could be pulled and executed automatically in the user's environment.
The setup example places a bearer token directly on the command line as an argument to npx. Command-line arguments are often exposed through shell history, process listings, logs, crash reports, and telemetry, which can leak reusable credentials to other local users or monitoring systems.
The recommended workflow examples use generic natural-language requests such as "Research rate limiting approaches...", "Save it to my... collection", and "Read the context from my... collection" without defining explicit invocation boundaries or exclusions. In a skill description, these broad phrases could overlap with ordinary conversation and make it unclear when the skill should activate versus when the user is just discussing work.