The Agent Skills Directory

[SAFE]: The skill uses official and local toolchain commands such as nim check and nimpretty. All operations are performed locally on the user's workspace without external network calls or credential access.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted local project files and tool outputs.
Ingestion points: Reads contents of .nim source files, .nimble configuration files, and stdout/stderr from the nim check command.
Boundary markers: The instructions do not specify explicit delimiters or isolation for the external data when presented back to the agent context.
Capability inventory: The skill executes shell commands (nim, nimpretty) to perform checks and repairs.
Sanitization: No specific sanitization or filtering is implemented for the data ingested from files or tools, though the workflow mandates a manual diff review by the user before any changes are committed.

codeguard-nim