check

Warn

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill resolves ecosystem-specific commands (build, test, and lint) through a "resolution ladder" where configuration files located inside the repository being analyzed (e.g., .claude/ecosystems/dotnet.yaml) are authoritative. This allows a potentially malicious repository to define arbitrary shell commands that the skill will execute during verification tasks.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection via repository configuration files. Adversaries can control the commands the agent executes by modifying the build-cmd, test-cmd, or check-cmd fields in the project's ecosystem YAML files.
  • Ingestion points: .claude/ecosystems/*.yaml files in the audited repository (SKILL.md).
  • Boundary markers: Absent. The skill does not implement delimiters or safety instructions to distinguish between its own logic and instructions provided in the repository configuration.
  • Capability inventory: Execution of arbitrary bash commands in the host environment (SKILL.md uses shell: bash).
  • Sanitization: Absent. Resolved command strings from the untrusted configuration files are executed directly as shell commands without validation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 30, 2026, 07:55 PM
Security Audit — agent-trust-hub — check