founder-brand
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests content from user-managed files and direct input to generate post drafts, creating a potential vector for instructions embedded in external data to influence agent behavior.
- Ingestion points: Instructions in
modes/content.mdandmodes/setup.mddescribe reading fromSOUL.md,brand-config.json, and direct user input for pillars, topics, and receipts. - Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for use when interpolating this external data into the generation prompts.
- Capability inventory: The skill possesses capabilities to execute specific local Python scripts via
Bash(check_setup.py,score_post.py) and perform file operations (Read,Write,Grep,Glob) within the project directory. - Sanitization: The skill instructions rely on internal voice rubrics and anti-patterns rather than structural sanitization or validation of the input data before processing.
Audit Metadata