Skills
skills/arenukvern/mcp_flutter/mcp-harness-repo-maintainer/Gen Agent Trust Hub

mcp-harness-repo-maintainer

Fail

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch an installation script from the author's GitHub repository: https://raw.githubusercontent.com/Arenukvern/skill_steward/main/install.sh.
  • [REMOTE_CODE_EXECUTION]: A recommended installation workflow involves piping the output of a curl command directly into bash, allowing for the execution of remote code without prior inspection. While the source is the skill author's repository, this remains a high-risk execution pattern.
  • [COMMAND_EXECUTION]: The skill utilizes several CLI commands from the steward toolset, including steward doctor, steward schema, steward probe, and steward benchmark, to perform repository audits and validations.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by ingesting and acting upon untrusted data from repository-local files.
  • Ingestion points: The agent reads configuration and logic from steward.yaml and scenario manifests in steward/scenarios/*.yaml.
  • Boundary markers: No explicit markers or instructions to ignore embedded commands are present in the instructions regarding configuration parsing.
  • Capability inventory: The steward CLI performs repository operations and executes actions defined in the configuration, which could include shell commands.
  • Sanitization: There is no mention of sanitization or strict schema validation for instructions that might be embedded in these files.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/Arenukvern/skill_steward/main/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 8, 2026, 01:03 PM
Security Audit — agent-trust-hub — mcp-harness-repo-maintainer