The Agent Skills Directory

[SAFE]: No malicious patterns or behaviors were detected across the analyzed files. The skill functions as a research framework and drafting aid for international law matters.
[NO_CODE]: The skill is composed exclusively of markdown files containing instructions, reference maps, and examples. It does not include any scripts (Python, JavaScript, shell), configuration files for automated execution, or binary data.
[EXTERNAL_DOWNLOADS]: The instructions direct the agent to use retrieval tools like web_fetch to access well-known and legitimate legal archives, such as stl-tsl.org (the official legacy archive of the Special Tribunal for Lebanon) and legal-tools.org (the ICC Legal Tools Database). These operations are well-scoped to the skill's purpose and target established institutional resources.
[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection because it is designed to ingest and process data from external websites and user-provided documents for auditing. However, the risk is mitigated by the skill's core "verification-first" methodology, which requires the agent to verify all content against multiple authoritative sources and follow a strict internal citation format.
Ingestion points: External websites (stl-tsl.org, legal-tools.org, un.org) and user-supplied drafts in audit mode.
Boundary markers: The skill mandates a specific "Standard workflow" and rigid citation formatting which serves as a process boundary, although no technical delimiters for the web_fetch output are explicitly defined.
Capability inventory: Uses network retrieval (web_fetch) to access public legal records.
Sanitization: The methodology relies on cross-verification (the "fallback ladder") against established primary sources to validate content accuracy and ignore unverified or malicious instructions in secondary data.

stl-lebanon-jeanne-sulzer