The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/analyze-all.py implements a fallback mechanism to dynamically load its sibling scripts (analyze-cso.py, analyze-compatibility.py, and check-char-budget.py) using importlib.util at runtime if standard module imports are unsuccessful. While limited to the skill's own installation directory, this uses computed paths for module loading.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, potentially untrusted skill files, which creates a vulnerability surface for indirect prompt injection.
Ingestion points: Skill files (SKILL.md, references, scripts) are read from the file system via user-provided paths in scripts like analyze-all.py and analyze-cso.py.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are employed within the scripts when reading and parsing the content of analyzed files.
Capability inventory: The skill utilizes Read, Write, and a scoped Bash tool limited to its internal scripts to perform its analysis.
Sanitization: The analysis scripts use regular expression matching to detect patterns but do not perform sanitization or escaping of the processed external text, allowing for potential prompt interpolation issues.

skill-crafting