social-media-context-sms
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to capture verbatim user input, specifically in the 'Example Posts' and 'Voice & Tone' sections, and store it in a persistent context file (
.agents/social-media-context-sms.md). This file is explicitly intended to be read by other agent skills, creating a surface where malicious instructions embedded in user-provided examples could influence downstream tasks. - Ingestion points: User input gathered in 'Step 3 — Gather information', particularly Section 7 (Example posts) and Section 3 (Voice & Tone).
- Boundary markers: The generated markdown file uses standard headers and horizontal rules as delimiters, but does not include explicit instructions for downstream agents to ignore embedded commands within the user-provided text.
- Capability inventory: The skill performs file-read and file-write operations to a specific path within the
.agents/directory. It does not perform network operations or execute arbitrary system commands. - Sanitization: The instructions explicitly mandate copying user posts 'verbatim' and capturing 'actual words... verbatim' to ensure voice matching, which precludes sanitization of potentially malicious instructions.
Audit Metadata