multi-phase-orchestrator
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it is designed to ingest and process data from external, untrusted sources like pull request comments to create automated work units.
- Ingestion points: The skill derives work units from user-provided plans, pull request comments, and review threads in
SKILL.md(Step 2). - Boundary markers: It implements explicit "Child Thread Packets" to define instructions and scope for delegated agents in
SKILL.md(Step 4). - Capability inventory: The agent has extensive capabilities including shell command execution, file modification, and git operations such as pushing and creating PRs in
SKILL.md(Steps 3, 6, 7, and 8). - Sanitization: The risk is mitigated by strict instruction to "Verify every finding before fixing it" and treating sub-agent outputs as mere "claims" until verified against live records in
SKILL.md(Core Rules). - [COMMAND_EXECUTION]: The skill facilitates the execution of developer tooling and repository scripts. It manages the risk of unauthorized or harmful execution through a "Validation Authority Contract" that specifically forbids sub-agents from running sensitive or resource-intensive commands (like Xcode/XCTest suites) without coordinator oversight.
Audit Metadata