The Agent Skills Directory

[COMMAND_EXECUTION]: The workflow runs shell commands like make -pn to inspect the build database and make -n for syntax validation. It also executes the help target of analyzed Makefiles to verify its functionality.
[REMOTE_CODE_EXECUTION]: The skill uses an external script makefile_dogfooder.py in the modules/plugin-dogfood-checks.md module for analysis, testing, and generation of targets (including applying changes with the --apply flag). This script is not included in the skill's file set.
[PROMPT_INJECTION]: The skill is vulnerable to indirect injection because it executes targets from user-provided Makefiles during validation. A malicious actor could provide a Makefile where the help or other audited targets contain malicious shell commands that the agent would then execute.
Ingestion points: Makefiles found via the **/Makefile glob.
Boundary markers: Absent; there are no instructions to verify target safety before execution.
Capability inventory: Subprocess calls via make and makefile_dogfooder.py.
Sanitization: Absent; the skill does not sanitize the contents of the Makefile before calling make on its targets.

makefile-review