openspec-ship-tickets
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches ticket content from external trackers (GitHub, Linear, Jira) and passes it verbatim to subagents as the primary instruction source. This creates a vector for malicious instructions in a ticket to influence subagent behavior.
- Ingestion points: External issue tracker (GitHub, Linear, Jira) issue bodies, project-local
tickets.mdfiles. - Boundary markers: Absent. The tracker body is treated as the source of truth for the subagent's scope.
- Capability inventory: Git operations (commit, push, rebase), file modification (via subagents), tracker comments/updates, and remote task delegation.
- Sanitization: Absent.
- [REMOTE_CODE_EXECUTION]: Uses platform-specific delegation tools to run sub-tasks in isolated environments.
- T3 Code: Uses
delegate_taskfor parallel execution in separate worktrees. - Amp: Uses the
ampCLI to spawn remote 'orb' threads. - [COMMAND_EXECUTION]: Executes shell commands for Git and tracker interaction.
- Git:
git fetch,git rebase,git push,git cherry-pick. - GitHub CLI:
gh issue view,gh issue comment,gh issue edit. - Jira/Linear: Interacts via MCP tools or CLI (e.g.,
jira issue view). - [EXTERNAL_DOWNLOADS]: Recommends installing dependencies and skills from the vendor's repository (
agoodway/GoodSkills) using thenpx skills addcommand.
Audit Metadata