implement
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external sources such as PR descriptions and issue reports which may contain malicious instructions.
- Ingestion points: Reads caller-supplied 'intent', 'acceptance', and 'scope' before execution (SKILL.md, Workflow Step 1).
- Boundary markers: The skill does not explicitly instruct the agent to use delimiters or 'ignore' prefixes when interpolating external intent into its processing loop.
- Capability inventory: The skill has the authority to write files, modify source code, and execute shell commands for building and testing (SKILL.md, generic-templates.md).
- Sanitization: No explicit input sanitization is defined, although the skill correctly identifies that sensitive environment tokens should be redacted from output logs (SKILL.md, Rule 4).
- [DYNAMIC_EXECUTION]: The skill generates and executes code and configuration dynamically as part of its scaffolding and repair functions.
- Evidence: Automatically generates project structures and components for multiple languages (Go, Python, Node, Rust) and executes verification commands like
go testorpytestimmediately after generation (generic-templates.md, Step 4). - [COMMAND_EXECUTION]: The skill relies on shell command execution for its primary functionality.
- Evidence: Instructions mandate running
git, theaoCLI, build tools, and linters to verify changes and collect evidence (SKILL.md, operations.md).
Audit Metadata