The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data via webhooks, which constitutes an indirect prompt injection surface. While the provided testing script only prints the data, real-world implementations using these instructions may be susceptible to malicious instructions embedded in call transcripts or metadata.
Ingestion points: The scripts/webhook_receiver.py script listens for HTTP POST requests and parses the JSON payload from external sources.
Boundary markers: The provided documentation and script do not implement specific boundary markers or instructions to ignore embedded commands within the payload.
Capability inventory: The script's capabilities are limited to parsing JSON and printing it to stdout. It does not perform any sensitive file operations, network requests (beyond listening), or command execution based on the received data.
Sanitization: The script performs basic JSON decoding but does not sanitize the contents of the payload (such as the transcript or recording URL) before printing.
[COMMAND_EXECUTION]: The skill documentation includes examples of using curl to interact with the Bolna API and instructions to execute a local Python script for testing. These are standard developer operations for the stated purpose of the skill.

setup-webhook