The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on a suite of Bash scripts (garden.sh, check-links.sh, check-naming.sh, check-index-coverage.sh, and analyze-search-gaps.sh) to perform file system analysis and health checks. The master script garden.sh manages these by ensuring internal scripts are executable using chmod +x and executing them as subprocesses within the local environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill logic involves parsing content from local files that could contain untrusted input.\n
Ingestion points: Data is read from SKILL.md and INDEX.md files using find, grep, and sed, and search queries are extracted from ~/.claude/skills/.search-log.jsonl using jq.\n
Boundary markers: The processing scripts do not implement explicit delimiters or markers to isolate the text being analyzed.\n
Capability inventory: The skill is limited to local file system reads and executing its own internal shell scripts. It possesses no network capabilities or permissions to write to sensitive system files.\n
Sanitization: Content is processed strictly as text strings for validation and reporting purposes and is not executed or interpreted as code instructions.

Gardening Skills Wiki