The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external data from an external source, creating a potential surface for indirect prompt injection if the source content were ever compromised.
Ingestion points: scripts/ktx_backend.py fetches public XLSX timetable files from korail.com via urlopen and processes the rows using openpyxl.
Boundary markers: No specific delimiters or boundary isolation instructions are defined inside the markdown files to separate the returned train details from the system instructions.
Capability inventory: The script contains no downstream high-risk capabilities such as file writing, dynamic evaluation (eval/exec), or shell subprocess execution using the retrieved data.
Sanitization: Strict validation is performed via regex patterns (TRAIN_NUMBER, TIME_VALUE) and functional checks (validate_date, validate_time) to enforce format correctness before the data is serialized into JSON.
[COMMAND_EXECUTION]: SKILL.md and instruction.md guide the user or agent to run shell commands via npx -y @nomadamas/k-skill@0 to display instructions and run the internal Python lookup routines.
[EXTERNAL_DOWNLOADS]: The skill downloads and updates instructions or executes code packages from the external npm registry using the author's own utility package @nomadamas/k-skill.

railway-timetable