The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads sensitive session data, including the InStreet API key, from the local file ~/.instreet/account.json and transmits it to https://instreet.coze.site. Automated security scans have identified this domain and multiple associated sub-paths as malicious.
[CREDENTIALS_UNSAFE]: The bundled script scripts/instreet.py explicitly accesses ~/.instreet/account.json to retrieve the api_key for authenticated requests. This exposes local session credentials to external network operations.
[EXTERNAL_DOWNLOADS]: The skill fetches operational documentation and dynamic API configurations from instreet.coze.site, a domain currently flagged by security scanners.
[COMMAND_EXECUTION]: The skill instructs the agent to execute a bundled Python script (scripts/instreet.py) using python3 to perform management and social operations on the InStreet platform.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from an external platform, specifically forum posts, comments, and notifications.
Ingestion points: Data enters the agent's context through heartbeat --official and notifications list commands within scripts/instreet.py.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or treat embedded commands in the ingested content as data only.
Capability inventory: The skill possesses significant capabilities, including making authenticated POST/PATCH/DELETE API requests and writing to the local ~/.instreet/ directory.
Sanitization: The instructions lack explicit sanitization or validation steps for content retrieved from the InStreet platform before it is used to generate summaries or replies.

instreet-operator