The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Connects to the official Twilio API at api.twilio.com to perform telephony operations.
[COMMAND_EXECUTION]: Uses curl to make HTTP requests and jq to extract specific fields from JSON responses.
[CREDENTIALS_UNSAFE]: Uses environment variables TWILIO_ACCOUNT_SID and TWILIO_AUTH_TOKEN for API authentication.
[PROMPT_INJECTION]: The skill ingests untrusted data (SMS message bodies and call details) from the Twilio API, which presents an indirect prompt injection surface.
Ingestion points: API responses from Messages.json and Calls.json in SKILL.md.
Boundary markers: No explicit delimiters are used in the prompt templates to separate message content from instructions.
Capability inventory: Includes network communication via curl and command execution via jq as described in SKILL.md.
Sanitization: Structured data is parsed using jq, but the textual content of message bodies is not sanitized before being returned to the agent context.

twilio