gs-openspec-ship-tickets

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection due to its handling of external tracker data. \n- Ingestion points: The orchestrator fetches ticket bodies, labels, and state from GitHub, Linear, or Jira (documented in SKILL.md and tracker-ops.md). \n- Boundary markers: Absent. The instructions in SKILL.md specify that the tracker body should be passed "verbatim" to subagents without delimiters or instructions to ignore embedded commands. \n- Capability inventory: Subagents triggered by this skill have extensive capabilities, including implementing code changes (openspec-apply-change), running security/code reviews (gs-review), and performing git commits and pushes (gs-git-commit, gs-pull-request). \n- Sanitization: Absent. No evidence of content filtering or escaping before data is interpolated into prompts. \n- [COMMAND_EXECUTION]: The skill executes multiple command-line tools to perform its orchestration. \n- Tooling: Regularly uses git, gh (GitHub CLI), jira-cli, amp, and npx. \n- Orchestration: In thread mode, it launches children using delegate_task or amp -ox, passing dynamically constructed prompts to manage the ticket pipeline.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 09:35 AM