save-harness
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data and persists it to the filesystem without adequate controls.\n
- Ingestion points: The skill ingests user-provided text meant to be saved as reusable assets (skills or prompts) in the
SKILL.mdfile.\n - Boundary markers: There are no specified delimiters or instructions to treat the saved content as non-executable data or to ignore embedded instructions.\n
- Capability inventory: The skill utilizes file-writing capabilities to store assets within the
~/Documents/harnessdirectory.\n - Sanitization: There is no implementation of sanitization, validation, or escaping to filter out potentially malicious commands from the data being saved.
Audit Metadata