court-payment-order-assistant
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied information, including debt details and evidence, to generate legal drafts and browser automation sequences, which represents a potential surface for indirect injection.
- Ingestion points: Information intake phase described in
instruction.md(Step 1), involving creditor, debtor, and claim facts. - Boundary markers: No explicit boundary markers or delimiters for user-provided facts are specified in the instructions.
- Capability inventory: The skill generates legal draft documents and interacts with a browser session via Chrome DevTools Protocol (CDP).
- Sanitization: Employs
validateIntakeandbuildRequiredQuestionsfunctions from the vendor's package for data validation. - [EXTERNAL_DOWNLOADS]: Fetches utility scripts and up-to-date instructions from the author's registry.
- Evidence: Use of
npx -y @nomadamas/k-skill@0inSKILL.mdto fetch and execute instructions and updates. - [REMOTE_CODE_EXECUTION]: Executes vendor-provided CLI tools at runtime to manage instructions and skill metadata.
- Evidence: Command execution patterns like
npx -y @nomadamas/k-skill@0 instructinSKILL.md. - [COMMAND_EXECUTION]: Provides shell commands for the agent to manage its environment and access current documentation.
- Evidence: Bash snippets in
SKILL.mdfor listing files and updating the skill via the vendor's CLI tool.
Audit Metadata