The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute repository-local maintenance scripts and version control checks.\n
Evidence: Instructions in SKILL.md and references/checklist.md to run node scripts/sync-plugin-skills.mjs and git diff --check to ensure consistency between the main repository and plugin skills.\n- [EXTERNAL_DOWNLOADS]: The skill references several external GitHub repositories and projects for layout research and as sources for copywriting guidelines.\n
Evidence: Links to projects such as Glow, Crush, Lazygit, and Atuin are used as architectural references. It also cites copy-editing rules adapted from open-source toolkits like marketingskills and agent-toolkit.\n
Note: These references are for informational and research purposes; no automated code downloads or piped executions from these sources are initiated.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes project-level documentation and manifests which could serve as ingestion points for untrusted data.\n
Ingestion points: Processes README.md, docs/*.md, and plugin manifests during review and editing tasks.\n
Boundary markers: Implements a strict 'line gate' and a mandatory checklist to verify all prose against verified facts rather than following potentially malicious or hallucinated text in the documents.\n
Capability inventory: Accesses shell commands for script synchronization and file diffing.\n
Sanitization: Explicitly requires the agent to verify all documentation claims, command flags, and terminal keys against the actual CLI help output and the repository's source code (internal/cli/, internal/config/).

repo-readme