pattern-of-life-from-socials
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external social media profiles, including metadata, post content, and user-generated comments. This creates a surface for indirect prompt injection attacks where malicious instructions could be embedded in the subject's public content to influence the agent's behavior.
- Ingestion points: As described in SKILL.md and reference/analytic-checklist.md, the agent is instructed to collect account metadata (IDs, bio text), network data (follower lists, mentions), and post content (text, timestamps, image metadata).
- Boundary markers: The provided instructions do not include specific delimiters or guidance for the agent to isolate ingested data from its own instructional context.
- Capability inventory: The skill references several tools that process the collected data, such as geolocate-from-pixels, secrets-in-file-metadata, read-deleted-pages, and graph-the-network.
- Sanitization: No explicit procedures for sanitizing or validating external content are defined before the data is analyzed or passed to other tools.
Audit Metadata