The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands and project-specific scripts to perform audits and modifications.
Evidence includes calls to rg, git, make, and pytest in SKILL.md, modules/deduplication-patterns.md, and modules/dependency-graph.md.
It executes makefile_dogfooder.py with several modes, including --apply, which may modify local project files (modules/plugin-dogfood-checks.md).
It executes citation_verifier.py as part of the validation workflow (SKILL.md).
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data from Makefiles which represents an attack surface for indirect prompt injection.
Ingestion points: Reads Makefile and *.mk files using rg and make -pn (SKILL.md, modules/dependency-graph.md).
Boundary markers: No explicit boundary markers or instructions to ignore embedded instructions in the analyzed files are defined.
Capability inventory: The skill can execute shell commands, run project scripts (makefile_dogfooder.py), and write to .review/findings.json (SKILL.md, modules/plugin-dogfood-checks.md).
Sanitization: There is no evidence of sanitization or validation of the content extracted from Makefiles before it is used in the review process.

makefile-review