The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's primary installation instructions in both SKILL.md and README.md recommend executing curl -fsSL https://parallel.ai/install.sh | bash. This is a critical security vulnerability as it executes a remote shell script from an untrusted external domain with the same privileges as the user's shell without any integrity or verification checks.
[COMMAND_EXECUTION]: The script scripts/generate_schematic.py uses subprocess.run to execute internal scripts. While arguments are passed as a list, the process involves forwarding raw user prompts into a secondary execution context, which increases the potential attack surface for command-related exploits.
[EXTERNAL_DOWNLOADS]: The skill relies on resources and APIs located at parallel.ai and openrouter.ai. These domains are not recognized as trusted organizations or well-known services within the defined scope, posing a significant supply chain and execution risk.
[PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection (Category 8) because it ingests and synthesizes untrusted data from third-party search tools and APIs into its final reports.
Ingestion points: External research results obtained via parallel-cli and the Perplexity/OpenRouter API responses.
Boundary markers: None identified; untrusted content is not isolated from instructions using delimiters or warning blocks.
Capability inventory: The skill is granted access to high-privilege tools including Bash, Read, Write, and Edit.
Sanitization: None; external data is parsed and directly formatted into markdown reports without any filtering or escaping mechanisms.

research-lookup