prompt-engineering
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is entirely markdown-based and acts as an instructional guide for the AI agent. It does not contain any executable scripts, binaries, or shell command invocations.
- [SAFE]: The skill performs automated reading of standard project configuration files (such as
package.json,go.mod,pyproject.toml, andCargo.toml) to detect the technology stack. These are non-sensitive metadata files typically used by coding assistants for context. - [PROMPT_INJECTION]: The skill defines a workflow for processing user-provided draft prompts, which introduces an indirect prompt injection surface. However, the risk is negligible as the skill is strictly advisory and lacks high-privilege capabilities such as network access or file-system modification.
- Ingestion points: Draft prompts provided by the user in the conversation.
- Boundary markers: None specified in the instructional logic.
- Capability inventory: Read-only access to standard project metadata and local documentation files (
CLAUDE.md,AGENTS.md). - Sanitization: No explicit sanitization or filtering of user-provided prompt content is defined.
Audit Metadata