skill-creator
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python and Bash scripts, such as
init_skill.pyandquick_validate.py, to automate the creation and validation of skill directories. These operations are performed on the local filesystem and align with the skill's intended purpose as a developer productivity tool. - [PROMPT_INJECTION]: The file
references/evals.yamlcontains several strings that match prompt injection patterns, including instructions to override system policies. These strings are contained within thepressure-prompt-injectionandpressure-skip-validationtest cases and are used to verify that generated skills can correctly identify and refuse adversarial inputs. - [SAFE]: The skill references absolute filesystem paths (e.g.,
/Users/jamiecraik/dev/configs/codex/agents) and external service URLs (e.g., GitHub Copilot MCP) in its documentation. These references are used for configuration examples and do not involve unauthorized data access or exfiltration.
Audit Metadata