The Agent Skills Directory

[SAFE]: The skill consists of markdown documentation and Python scripts designed for compliance assessment. The scripts (ai_act_readiness_score.py and gpai_obligation_checker.py) rely exclusively on the Python standard library (argparse, json, dataclasses, pathlib) and do not perform any network operations or access sensitive system files beyond the user-provided configuration.
[INDIRECT_PROMPT_INJECTION]: The assessment tools ingest user-provided YAML files (ai-system.yaml, model.yaml) to generate readiness reports.
Ingestion points: The scripts read text content from files specified via the --config and --model CLI arguments.
Boundary markers: The scripts output results in Markdown or JSON format, but do not explicitly wrap these in protective boundary markers for the agent.
Capability inventory: The scripts are restricted to local file read/write (via Path) and standard output. No network or subprocess capabilities are present.
Sanitization: The scripts use a custom recursive parser for YAML content that handles basic data types and structure without dangerous evaluation (no eval() or exec() on input data). The risk is minimal as the scripts only perform scoring logic and do not interpolate untrusted data into instructions or dangerous tool calls.

ai-act-readiness