The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructions and scripts interact with local files containing sensitive information.
The SKILL.md quick start guide uses cat ~/.config/elevenlabs/api_key to load an API key into an environment variable.
The documentation references config/api_keys.json as a storage location for Telnyx and Twilio authentication details.
[COMMAND_EXECUTION]: The documentation includes usage examples that pipe remote data into a command interpreter.
SKILL.md contains an example command: curl -s "https://api.elevenlabs.io/v1/voices" ... | python3 -c "...", which executes inline Python code on the results of a network request to ElevenLabs.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection via dynamic variables.
Ingestion points: The scripts/call.sh script accepts external data through the --vars argument.
Boundary markers: No specific boundary markers or instructions are provided to the AI to ignore potentially malicious instructions within these variables.
Capability inventory: The skill possesses the capability to initiate outbound phone calls (scripts/call.sh) and retrieve conversation transcripts (scripts/conversation.sh).
Sanitization: While the script uses jq for JSON structure integrity, it does not sanitize the content of the variables against prompt-based attacks.

elevenlabs-calls