The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides several bash scripts (scripts/transcript.sh, scripts/scrape.sh, scripts/poll-job.sh) that use curl to perform network operations and jq to process data. These scripts are invoked to communicate with the vendor's API at api.supadata.ai.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to fetch and process untrusted external data (video transcripts, social media metadata, and web content).
Ingestion points: External data is introduced into the agent's context through the output of scripts/transcript.sh, scripts/scrape.sh, and scripts/poll-job.sh.
Boundary markers: The instructions do not specify any delimiters or warnings to the agent to distinguish between the fetched content and instructions, increasing the likelihood of the agent following malicious instructions embedded in a video's captions or a website's text.
Capability inventory: The skill possesses the capability to execute shell commands and perform network requests to external domains.
Sanitization: The skill does not implement any sanitization or validation of the retrieved external content before presenting it to the agent.

supadata