x-intelligence-automation
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from X/Twitter (tweets, user profiles, trends) which serves as an ingestion point for external content that may contain instructions designed to influence the agent's behavior.
- Ingestion points: External data is fetched from the Twitter API via the AIsa relay in
scripts/twitter_client.py. - Boundary markers: Guidelines in
references/engage_twitter.mdandreferences/post_twitter.mdinstruct the agent to confirm user targets and not fabricate information, though the scripts do not implement strict technical delimiters for the external text. - Capability inventory: The skill possesses capabilities to perform network requests to
api.aisa.one, read local workspace files for media uploads viascripts/twitter_oauth_client.py, and execute account actions such as posting or following. - Sanitization: The
scripts/twitter_oauth_client.pyclient validates that media uploads are existing files and restricts the content to image or video MIME types before transmission to the relay.
Audit Metadata