The Agent Skills Directory

[PROMPT_INJECTION] (LOW): The skill is susceptible to indirect prompt injection because it processes untrusted data from X (Twitter).
Ingestion points: Data enters the context via /x search, /x profile, and /x thread commands which retrieve public tweets.
Boundary markers: Absent. There are no instructions provided to the agent to treat retrieved tweet content as data only or to ignore embedded instructions.
Capability inventory: The skill appears limited to data retrieval and display (read-only), which mitigates the impact of potential injections.
Sanitization: No sanitization or filtering logic is defined in the skill documentation.
[DATA_EXFILTRATION] (LOW): The skill requires network communication with a non-whitelisted domain.
Evidence: The skill documentation references and utilizes the Composio API (https://composio.dev). While a legitimate integration platform, it is not included in the 'Trusted External Sources' whitelist, categorizing the network activity as a low-severity finding.
[CREDENTIALS_UNSAFE] (SAFE): The skill requires COMPOSIO_API_KEY and COMPOSIO_CONNECTION_ID. These are correctly defined as environment variable requirements rather than being hardcoded in the skill text.

x-research