work-on-issue

Pass

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • Command Execution: The skill automates development workflows by executing local shell scripts (worktree-create.sh and tdd-state.sh) located within the .claude/ directory. These operations are intended for managing Git worktrees and the state of TDD sessions.
  • Indirect Prompt Injection Surface: The skill ingests data from external GitHub issues to derive identifiers for shell command arguments, such as the short-description for worktree names.
  • Ingestion points: External content is fetched from GitHub issues via the issue-worker sub-agent specified in SKILL.md.
  • Boundary markers: The sub-agent prompt does not currently utilize specific boundary markers or delimiters for the external content.
  • Capability inventory: The skill possesses the capability to execute shell commands and write todos based on the ingested content.
  • Sanitization: The workflow relies on the sub-agent to extract a concise description, which provides a natural language filter, though it does not explicitly implement shell-escaping for the resulting strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 8, 2026, 04:16 AM
Security Audit — agent-trust-hub — work-on-issue