The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository (architectural records and source code), creating a surface for indirect prompt injection. Maliciously crafted ADRs or code comments could attempt to influence the agent's behavior or output.
Ingestion points: The skill reads files in handbook/engineering/decisions/ and code changes identified via git diff.
Boundary markers: None. The skill does not define specific delimiters or instructions to ignore embedded content when reading files.
Capability inventory: The skill is limited to reading local files and executing git and ls commands; it lacks network or write capabilities.
Sanitization: None specified for the content read from files.
[COMMAND_EXECUTION]: The skill uses local shell commands to identify project state and file lists.
Evidence: Uses git diff --name-only and ls handbook/engineering/decisions/[0-9]*.mdx to scope the analysis. These commands are limited to local repository metadata and do not include user-controlled arguments that could lead to command injection.

adr-check