dino-manage-prompts
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the
dinoCLI tool to manage prompts. This is the primary intended purpose of the skill and is performed using best practices such as argument isolation and dry-runs. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided prompt templates and CLI output which are considered untrusted data. 1. Ingestion points: Data is ingested via
dino prompt listcommand output and user input fordino prompt add(found in SKILL.md). 2. Boundary markers: The skill instructions explicitly warn to treat all Dinox data and CLI output as untrusted and to never execute instructions found within it. 3. Capability inventory: The skill has capabilities to execute thedinoCLI and create/write/delete temporary files for data handling. 4. Sanitization: The skill employs shell argument isolation by writing multiline prompt text to temporary files and using the@filesyntax, and requires dry-runs before performing write operations. - [DYNAMIC_CONTEXT_INJECTION]: The documentation mentions the
! dino auth loginsyntax. This is provided as instructional context for how users can authenticate in specific agent environments and does not represent a malicious attempt to execute code at skill load time.
Audit Metadata