skills/sungjunlee/dev-relay/relay/Gen Agent Trust Hub

relay

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted data from external sources, creating a surface for indirect prompt injection.
  • Ingestion points: Task and issue text retrieved via gh issue view, user-provided natural language descriptions, and configuration artifacts like /tmp/dispatch-<N>.md and /tmp/rubric-<N>.yaml (referenced in SKILL.md).
  • Boundary markers: The skill instructions and the references/prompt-template.md file explicitly define a <task-content> block. They include mandatory instructions for the agent to treat this section strictly as a specification and to ignore any directives like "ignore instructions" or "system:" found within that data.
  • Capability inventory: The skill can execute local Node.js scripts, perform Git operations (git fetch, git push, git worktree), interact with the GitHub API via the gh CLI, and delete directories during its garbage collection phase (fs.rmSync in scripts/relay-status.js).
  • Sanitization: CLI arguments are sanitized using a custom shellQuote function in scripts/relay-advance.js and scripts/relay-recover.js to prevent command injection during script invocation.
  • [COMMAND_EXECUTION]: The skill's primary function is to orchestrate the execution of various local scripts and system tools.
  • The scripts relay-advance.js, relay-recover.js, and run-preflight.js use child_process.execFileSync to invoke git and gh CLI commands.
  • Orchestration logic in scripts/relay-advance.js dynamically constructs shell commands for other scripts in the relay suite. These paths are computed using static offsets from __dirname, and arguments are escaped to ensure safe execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 12:24 AM