Uniswap CCA
Warn
Audited by Gen Agent Trust Hub on Jul 5, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill initiates installation via the command
npx skills add Uniswap/uniswap-ai. This process fetches additional configuration and scripts from the official Uniswap organization repository, which is a verified and trusted source. - [PROMPT_INJECTION]: The skill metadata (SKILL.md and catalog.json) contains misleading attribution by claiming the skill is 'Provided by Uniswap' and referencing official Uniswap domains, while the actual skill author is identified as 'BankrBot'. This deception can cause users and agents to misjudge the safety and origin of the code.
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it interpolates user-supplied data into generated Foundry deployment scripts and Python configuration files.
- Ingestion points: Auction parameters like token addresses, prices, and supply curves are accepted as inputs for the
/configuratortool. - Boundary markers: There are no explicit delimiters or instructions to the agent to treat these parameters as raw data only.
- Capability inventory: The
/deployertool described incatalog.jsongenerates and prepares Foundry scripts and Python-based servers for execution. - Sanitization: No validation or sanitization routines are identified for ensuring the integrity of the user-provided parameters before they are used in script generation.
Audit Metadata