The Agent Skills Directory

[SAFE]: The skill implements statistical rigor gates and pre-registration of metrics, which are industry best practices for preventing analytical bias and p-hacking.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute Python scripts for data processing and metric calculation. These scripts are constructed based on safe templates provided in the reference files (compute-examples.md and rigor-gates.md).
[DATA_EXFILTRATION]: No network-enabled tools are used, and no instructions for remote data transmission were found. The skill operates entirely on local datasets and generates local markdown artifacts.
[PROMPT_INJECTION]: The instructions do not contain any patterns designed to bypass system constraints or safety filters. The use of natural language is strictly instructional and focused on data science methodology.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface Analysis: 1. Ingestion points: User-provided CSV and JSON files processed in Phase 3. 2. Boundary markers: The skill instructions focus on structural parsing rather than content isolation. 3. Capability inventory: The agent has access to Bash for computation and Write for file storage. 4. Sanitization: No explicit sanitization of data strings is mentioned. This represents a standard surface for data processing tools and is mitigated by the skill's focus on statistical aggregation rather than string interpretation.

data-analysis