The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides scripts (lab-probe.sh and lab-probe.ps1) designed to probe system capabilities such as RAM, CPU, and Docker status. These scripts execute local system utilities (docker info, df, sysctl, etc.) to determine suitability for optional Docker-based documentation tools.
[EXTERNAL_DOWNLOADS]: The documentation suggests the installation of specific NPM packages (@carrilloapps/docgraph and skill-rules) and Docker images (from davidanson, jdkato, and lycheeverse) for markdown linting and link checking. These are presented as opt-in suggestions with pinned versions and integrity hashes.
[INDIRECT_PROMPT_INJECTION]: As the skill is designed to process and index project documentation, it defines an untrusted input boundary, explicitly instructing the agent to treat content from project files as data and never as instructions, mitigating potential prompt injection via analyzed files.

ai-rules