The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests local C source files and executes shell commands based on their paths, which presents a surface for indirect prompt injection.\n
Ingestion points: The skill identifies and scans files with .c and .h extensions using the find command and clang-tidy.\n
Boundary markers: No explicit markers are defined to separate file data or paths from the command instructions during shell execution.\n
Capability inventory: The skill uses bash, find, xargs, clang-tidy, and clang-format to perform its operations, as seen in SKILL.md and references/tooling/toolchain.md.\n
Sanitization: There is no evidence of explicit sanitization or escaping of file paths before they are passed to the shell, although the skill instructions mention using authorized paths.

codeguard-c