The Agent Skills Directory

[SAFE]: The skill interacts with the redfox.hk domain to retrieve video metadata. This domain is owned by the author, redfox-data, and the communication is essential for the skill's primary search functionality.
[DATA_EXFILTRATION]: The script search_wechat_channels.py accesses the REDFOX_API_KEY environment variable to authenticate API requests. This is a standard implementation for services requiring authentication and does not represent an unauthorized exfiltration of sensitive system credentials.
[PROMPT_INJECTION]: The skill processes content such as video titles and descriptions fetched from a remote API. This represents a potential surface for indirect prompt injection if the remote content contains malicious instructions.
Ingestion points: External metadata is ingested in scripts/search_wechat_channels.py through the API response.
Boundary markers: There are no explicit delimiters defined in the agent instructions to isolate the external content.
Capability inventory: The agent can execute the search script, write JSON output to files, and manage scheduled tasks via qoder_cron.
Sanitization: The agent is instructed to format the output into a structured Markdown table, which helps constrain the influence of the ingested data.
[COMMAND_EXECUTION]: The skill uses qoder_cron to schedule periodic searches. This persistence mechanism is explicitly tied to the skill's 'subscription' feature, which requires user confirmation before activation.

wechat-channels-crawler