The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation section provides a command to execute remote code using npx agentic-awesome-skills --claude. Running npx without a specific version or verification of the source allows for the execution of arbitrary scripts from the npm registry, posing a supply-chain risk since the author and package are not from a verified or trusted vendor.
[EXTERNAL_DOWNLOADS]: The skill references a non-standard installation path via the agentic-awesome-skills package. This involves downloading and executing content from an external community repository that has not been vetted for safety.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and analyze untrusted source code provided by the user or found in a codebase.
Ingestion points: Files passed to the tool, such as src/services/PaymentService.ts or entire modules during architectural reviews.
Boundary markers: None. The instructions do not specify delimiters or tell the agent to ignore instructions embedded within comments or strings in the reviewed code.
Capability inventory: The skill uses advanced language models (claude, codex, cursor, gemini) to perform analysis, which could be manipulated to return biased results or ignore design smells if the input code contains adversarial prompts.
Sanitization: There is no evidence of sanitization or filtering of the input source code before it is processed by the AI models.

brooks-lint