The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the user or agent to execute local Python scripts (scripts/akta.py and scripts/bramka_lokatorow.py) to process documents and verify the accuracy of the resulting chronology.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted legal text files (case records) which presents an attack surface for instructions embedded in the data to influence the agent's output.
Ingestion points: Text files located in the akta-tekst directory.
Boundary markers: The instructions explicitly require the agent to use "kotwice" (anchors) consisting of the file name, page number, and literal quotes.
Capability inventory: The skill uses a dedicated script (bramka_lokatorow.py) that performs a deterministic, non-LLM check to ensure every claim is backed by a verifiable quote in the source files.
Sanitization: The export functionality in bramka_lokatorow.py includes a security measure to neutralize potential spreadsheet formula injections (e.g., =, +, -, @) by prepending a single quote to cells in the CSV output.

chronologia-stanowiska-pl