software-project-starter

Pass

Audited by Gen Agent Trust Hub on Nov 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute various system commands for project setup and validation.
  • Evidence: Commands include npx skills add, npx skills update, git config, and docker build.
  • A shell script is provided in references/docker.md for inspecting Docker build contexts using mktemp, docker build, find, and sed to verify file inclusion rules.
  • [EXTERNAL_DOWNLOADS]: The skill downloads and executes code or configurations from third-party sources.
  • Evidence: The skill uses npx skills add <source> to fetch and install additional functionality from GitHub repositories, specifically mentioning samber/cc-skills-golang in references/go.md.
  • It recommends the use of the immanuwell/dockerfile-roast GitHub Action for Dockerfile linting in references/docker.md.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing untrusted data from the local project environment.
  • Ingestion points: The skill explicitly reads project files and user-provided instructions in the target directory to "frame the project" and "inspect a user-named starter" (SKILL.md).
  • Boundary markers: There are no instructions to use delimiters or ignore instructions embedded within the processed project files.
  • Capability inventory: The skill has extensive capabilities, including file system modification, command execution via npx and git, and container management via docker.
  • Sanitization: No sanitization or validation of the content read from the project directory is performed before it influences the agent's behavior and planning logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 1, 2026, 10:26 AM