shemic-dever

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads the 'Dever' framework and CLI tool from the vendor's repository (github.com/shemic/dever) during project initialization using the go get command.
  • [REMOTE_CODE_EXECUTION]: The skill executes the dever CLI tool directly from its remote repository using go run github.com/shemic/dever/cmd/dever@main. The use of the @main tag means the code being executed is determined at runtime by the latest state of the remote repository.
  • [COMMAND_EXECUTION]: The skill invokes local shell scripts (scripts/boot.sh and scripts/module.sh) to automate the scaffolding of project directories and the generation of Go source code files.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it processes user-provided 'requirement descriptions' to generate code and configuration.
  • Ingestion points: User requirement descriptions (e.g., module names, resource names, and interface actions) provided in the agent's prompt context.
  • Boundary markers: No explicit delimiters or instructions are used to separate untrusted user inputs from the code generation templates.
  • Capability inventory: The skill possesses capabilities for writing to the local filesystem (cat > file), fetching remote modules (go get), and executing code (go run).
  • Sanitization: The provided scripts perform basic formatting (such as camel-case conversion) on user inputs but do not include comprehensive validation or sanitization of the content being written into source files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 05:05 AM