The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides an interface to ingest untrusted data that is then processed and transmitted to an external system, creating a potential surface for indirect prompt injection.
Ingestion points: The create and lookup commands in scripts/create.py ingest data from the --data and term arguments respectively.
Boundary markers: No specific boundary markers or instructions to ignore embedded instructions are implemented within the skill scripts or instructions.
Capability inventory: The scripts/create.py script possesses network capabilities to perform POST requests via urllib.request.urlopen.
Sanitization: Input data is parsed as JSON but does not undergo content-level sanitization before being included in the API request payload.
[DATA_EXFILTRATION]: The skill accesses local credentials and transmits them to the Konecty API to perform authorized actions. This is documented behavior consistent with the vendor's intended functionality.
Evidence: scripts/create.py reads sensitive configuration from ~/.konecty/.env and ~/.konecty/credentials.
Context: The data access and transmission are limited to the Konecty platform's own endpoints as defined in the user's configuration.

konecty-create