The Agent Skills Directory

[SAFE]: The skill consists of instructional text providing guidance on data harmonization and merging techniques for datasets from Our World In Data. It does not contain executable code, scripts, or active components.
[INDIRECT_PROMPT_INJECTION]: The instructions describe workflows that involve ingesting data from external, untrusted sources, which establishes a potential surface for indirect prompt injection. 1. Ingestion points: External data sources and regional documentation referenced in SKILL.md. 2. Boundary markers: Absent; no delimiters are defined in the instructions. 3. Capability inventory: Bash(duckdb:*) is permitted for data manipulation. 4. Sanitization: Absent; the instructions do not provide specific patterns for validating external input.

joining-data