The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md (Companion check) direct the agent to execute a shell command (ls) on user-level directories such as ~/.claude/skills/, ~/.agent/skills/, and others to identify existing installations of recommended skills.
[EXTERNAL_DOWNLOADS]: Both the README.md and SKILL.md (Companion check) suggest the installation of the skill and its dependencies from the AbsolutelySkilled GitHub repository using the npx skills add command.
[REMOTE_CODE_EXECUTION]: The use of the npx skills add command for installation and skill management involves downloading and executing logic from a remote source (github.com/AbsolutelySkilled/AbsolutelySkilled).
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core functionality involves reading and processing untrusted data from the local codebase to inform its reasoning.
Ingestion points: The 'Deep Context Scan' (Phase 1) and 'Codebase-First Intelligence' (Phase 2) in SKILL.md read README.md, CLAUDE.md, CONTRIBUTING.md, and other files in the docs/ directory.
Boundary markers: No explicit delimiters or boundary instructions are provided in the skill instructions to prevent the agent from following directions embedded within these files.
Capability inventory: The skill has the capability to read any file, search the codebase, and initiate a sub-agent review process (Phase 8).
Sanitization: There is no mention of sanitizing or escaping the content retrieved from the codebase before it is integrated into the agent's context.

absolute-brainstorm