The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/biz_health_check.py dynamically loads and executes Python modules from sibling directories using importlib.util. The module paths are computed relative to the script's location (_REPO_ROOT / skill_dir / "scripts" / filename). This facilitates a modular architecture where the composite skill reuses logic from independent sub-skills.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a data aggregator for business-related information, creating a surface for potential indirect prompt injection through processed external content.
Ingestion points: Business registration numbers (b_no), company names (name), and regional parameters (region) provided to the run function in scripts/biz_health_check.py.
Boundary markers: The skill output includes a clear disclaimer field stating that the report is a collection of facts without interpretation, intended for the user's evaluation.
Capability inventory: The script performs dynamic module loading and passes configuration (like base_url) to modules that perform network-based queries.
Sanitization: The _normalize_b_no function uses regular expressions (re.sub and re.fullmatch) to strictly validate that the registration number is a 10-digit numeric string.
[EXTERNAL_DOWNLOADS]: The skill uses a vendor-owned CLI tool, @nomadamas/k-skill, which is fetched and executed via npx as described in SKILL.md and instruction.md. This tool is used to synchronize instructions and manage the skill's execution environment.

biz-health-check