The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from the Seoul Open Data Plaza via a proxy, which serves as an ingestion point for untrusted content.
Ingestion points: Real-time bike station data is fetched in scripts/seoul_bike.py via the fetch_json function calling the /v1/seoul-bike/nearby and /v1/seoul-bike/realtime endpoints.
Boundary markers: The script output lacks explicit delimiters or instructions to the agent to disregard any commands that might be found within the station_name or other data fields.
Capability inventory: The skill uses npx to execute CLI commands and urllib to perform network operations.
Sanitization: While the script performs type conversion for numeric fields, it does not sanitize or escape text strings retrieved from the external API before presenting them to the agent.
[EXTERNAL_DOWNLOADS]: The skill relies on an external CLI tool distributed via npm.
Evidence: Instructions in SKILL.md and instruction.md direct the agent to run npx -y @nomadamas/k-skill@0, which downloads and executes the package from the npm registry. This is the primary mechanism for skill execution and updates.
[COMMAND_EXECUTION]: The skill uses shell commands to provide its core functionality.
Evidence: The entrypoint defined in instruction.md uses npx to execute the local Python script scripts/seoul_bike.py. This involves passing arguments from the agent to the shell, which is the intended method for this skill's CLI-based operation.

seoul-bike