The Agent Skills Directory

[External Downloads & Remote Code Execution] (HIGH): The skill documentation encourages the use of npx -y @grimoirelabs/cli. This command downloads and executes code directly from the npm registry. Because '@grimoirelabs' is not a trusted organization according to defined security standards, this constitutes a high-risk external dependency that could facilitate a supply chain attack or arbitrary code execution.
[Command Execution] (MEDIUM): The skill is designed to run several CLI-based commands, including bun run and direct binary execution. This provides a mechanism for system-level interaction that, while intended for the skill's functionality, lacks restricted sandboxing.
[Indirect Prompt Injection] (LOW): The commands for tokens and pools accept external URLs via the --source and --endpoint flags. This allows the skill to process content from attacker-controlled servers.
Ingestion points: External data is ingested through the --source and --endpoint arguments in SKILL.md.
Boundary markers: Absent. The skill does not provide delimiters or instructions for the AI to treat the resulting data as untrusted.
Capability inventory: The skill can execute shell commands, run Node.js scripts, and fetch data from the network.
Sanitization: None detected. The CLI outputs are processed as raw data without visible validation or escaping.

grimoire-uniswap