The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically through actions like comments (listing post comments), messages (listing direct messages), and reviews (fetching Google Business reviews). This content is processed by the agent and could contain instructions intended to override its behavior.
Ingestion points: API responses for comments, messages, reviews, and brand actions in scripts/x2c_social.py.
Boundary markers: None detected; the instructions do not advise the agent to treat this data as untrusted or separate it from instructions.
Capability inventory: The agent can read and upload local files, make network requests to the X2C API, and post content to various social media platforms.
Sanitization: No sanitization of ingested content was observed before processing by the agent.
[DATA_EXFILTRATION]: The publish, auto-schedule, and upload actions in scripts/x2c_social.py allow for the reading and transmission of local files to the vendor's API endpoint. While this is a functional requirement for sharing media, it provides a capability that could be exploited to exfiltrate sensitive local files if the agent is manipulated into selecting inappropriate file paths.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to ukbyfmmerxhlghlhbxtz.supabase.co, a domain associated with a well-known cloud service provider, to interact with the X2C API.

x2c-socialposter