The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill makes network requests to https://redfox.hk to submit video URLs and poll for transcription results. This domain belongs to the skill's author (redfox-data) and is used for core functionality.
[COMMAND_EXECUTION]: The skill includes a Python script scripts/fetch_audio_text.py designed to automate interaction with the vendor API. The script uses standard libraries and performs basic input validation on the provided URL.
[CREDENTIALS_UNSAFE]: The skill implements secure credential management by requiring the REDFOX_API_KEY to be set as an environment variable, preventing sensitive keys from being hardcoded in the skill files.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes external content (video transcripts) which are then summarized by the agent.
Ingestion points: User-provided video URLs are processed into text via scripts/fetch_audio_text.py.
Boundary markers: Absent; the workflow instructs the agent to summarize the text directly without specific isolation tags.
Capability inventory: The skill can write results to local markdown files and communicate with the vendor API.
Sanitization: The script does not perform content filtering on the returned transcript, relying on the underlying LLM's safety filters.
[DATA_EXFILTRATION]: No evidence of unauthorized data transmission was found. The only data sent externally is the video URL provided by the user, which is directed to the intended vendor service.

video-link-extract