skills/wkentaro/skills/clean-branch/Gen Agent Trust Hub

clean-branch

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill identifies feature scope and intended behavior by inspecting external sources such as GitHub issues or Pull Requests. This creates a surface for indirect prompt injection where an attacker could embed malicious instructions in an issue or PR to influence the agent's git operations.
  • Ingestion points: Data is pulled from "relevant issue or PR" in the Identify the feature and base section of SKILL.md.
  • Boundary markers: There are no instructions to use delimiters or ignore potentially malicious commands embedded in the issue/PR text.
  • Capability inventory: The skill uses git worktree and other git commands across all sections of SKILL.md to create branches and modify the repository structure.
  • Sanitization: The instructions do not include any validation or sanitization steps for the data retrieved from issues or PRs.
  • [COMMAND_EXECUTION]: The skill performs repository modifications using git commands, specifically creating new branches and worktrees. While this is the intended functionality, it represents the primary capability set available if a prompt injection occurs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 02:23 AM
Security Audit — agent-trust-hub — clean-branch