anoma-intents
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [NO_CODE] (INFO): The skill consists exclusively of documentation (README.md) and contains no executable source code or automation scripts.
- [COMMAND_EXECUTION] (LOW): The skill documentation describes the use of
juvixandanoma-clifor compiling and submitting intents. While these are shell-based tools, no logic for their automated execution is present in the provided file. - [EXTERNAL_DOWNLOADS] (LOW): The documentation references external binaries
juvixandanoma-cli. These are required dependencies for the workflows described but are not automatically downloaded by the skill itself. - [PROMPT_INJECTION] (LOW): Category 8 (Indirect Prompt Injection) vulnerability surface identified. 1. Ingestion points: User-provided intents in Juvix or Clojure format. 2. Boundary markers: Absent. 3. Capability inventory: Shell access for compilation and submission tools. 4. Sanitization: Absent. The skill encourages the agent to translate user 'wants' into executable actions, which could be exploited if the user input contains embedded instructions.
Audit Metadata