The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill requires the agent to read and edit documentation files (Ingestion points: docs/decisions/ and templates). It lacks boundary markers to separate external content from instructions. While capabilities are limited to file system modification and local git queries, malicious content within existing records or templates could influence the agent's reasoning during the documentation process.
[Command Execution] (LOW): The instructions direct the agent to execute git log --format="%ai" <commit> -1 to verify dates. This involves executing a shell command with a variable commit parameter, which constitutes a standard but exploitable command execution surface if the parameter originates from an untrusted source.

decisions