task-coordinator
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPERSISTENCECOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for workflow orchestration where sub-agents are spawned based on external data such as reviewer lists and task metadata. This role as a controller creates an attack surface where instructions embedded in workflow data could potentially influence sub-agent behavior.
- Ingestion points: Workflow definitions, parallel reviewer lists, and task metadata fields (SKILL.md).
- Boundary markers: No specific delimiters or safety instructions are defined in the orchestration templates to isolate untrusted task data.
- Capability inventory: The skill utilizes sub-agent spawning (Task()), file system writing (ledger checkpoints), and environment variable modification.
- Sanitization: The documentation does not include explicit validation or sanitization steps for data passed to sub-agents.
- [PERSISTENCE]: The skill incorporates state management techniques to enable crash recovery and cross-session work. This includes writing workflow state to a local file (.agents/workflow-ledger.yml) and storing task identifiers in environment variables (CLAUDE_CODE_TASK_LIST_ID).
- [COMMAND_EXECUTION]: Orchestrates the execution of multiple sub-agents using the platform's native Task API and handles file operations for maintaining the workflow ledger.
Audit Metadata