The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external CSV files placed in the inbox/ directory to track social media metrics. This external input could potentially be used as a vector for indirect prompt injection to influence the agent's analysis or recommendations.
Ingestion points: scripts/ingest-csv.py reads user-supplied CSV exports.
Boundary markers: No specific boundary markers or "ignore embedded instructions" warnings are used when processing the ingested metrics.
Capability inventory: The skill is capable of executing shell commands and modifying X (Twitter) profile information through the xurl utility.
Sanitization: The ingest-csv.py script validates numeric inputs (e.g., visits, follows) but does not sanitize potential text payloads in headers or other non-numeric fields.
[COMMAND_EXECUTION]: The skill's operational logic involves executing multiple Python scripts and the xurl CLI tool to perform data ingestion, experiment evaluation, and account management.
[DYNAMIC_EXECUTION]: The scripts/scoreboard.py script uses importlib.util to dynamically load the scripts/evaluate.py module during execution. While this targets a local file, dynamic loading is a complex execution pattern that increases the surface area for logic manipulation.

wjs-x-increasing-follower