The Agent Skills Directory

[SAFE]: The skill utilizes a deterministic Python script (disclosure_builder.py) to manage complex policy logic and statement drafting. This script is strictly limited to the Python standard library, requires no network access, and does not use dangerous functions such as eval(), exec(), or pickle.loads().
[SAFE]: All external URLs referenced in the documentation and reference files point to official, verified domains of legitimate academic publishers (Elsevier, Springer Nature, Wiley), government funding agencies (NIH, NSF, UKRI, TÜBİTAK), and regulatory bodies (European Union, YÖK).
[SAFE]: The skill includes explicit defensive instructions that direct the agent to decline user requests to hide or understate AI usage, preventing the agent from being manipulated into generating deceptive compliance documentation.
[COMMAND_EXECUTION]: The skill uses the Bash tool to run its local disclosure builder script. The execution is deterministic and operates on local files provided within the skill package, representing a safe use of the tool for its intended primary purpose.
[DATA_EXFILTRATION]: No exfiltration vectors were identified. The skill does not access sensitive local files (like .ssh or .aws credentials) and does not perform network operations to non-whitelisted or suspicious domains.

alterlab-ai-use-disclosure