The Agent Skills Directory

[SAFE]: All external URLs and data sources referenced in the skill point to well-known, trusted international organizations, including the IMF, OECD, World Bank, and V-Dem Institute.
[NO_CODE]: The skill consists entirely of instructional markdown and configuration. It does not include any executable scripts, binaries, or automated installation commands.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for ingesting and analyzing external political and economic data. While this presents an indirect prompt injection surface, the risk is mitigated by the skill's enforcement of structured methodologies and the use of authoritative, peer-reviewed data sources.
Ingestion points: The agent is instructed to use external datasets (IMF, OECD) and comparative news contexts.
Boundary markers: The skill requires "specific citations" and "evidence-based" analysis, which serves as a functional boundary for checking source validity.
Capability inventory: The skill does not request tool access or the ability to execute code, perform network writes, or modify the file system.
Sanitization: The methodology requires the agent to "read ALL output back completely" and "systematically improve every section," facilitating the identification and correction of any anomalous or injected content during the iterative process.
[SAFE]: No obfuscation (e.g., Base64, zero-width characters) or persistence mechanisms were detected. The "AI FIRST Quality Principle" is a stylistic and quality-focused instruction rather than a prompt injection attempt to bypass safety constraints.

comparative-politics-reporting