agent-loop-orchestrator
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill triages and dispatches autonomous fixes based on repository issues and pull requests, which are untrusted external data sources.
- Ingestion points: Data from issues and pull requests is ingested via the triage sub-skill (
../agent-loop-triage/SKILL.md) and passed to subagents. - Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the untrusted issue/PR content.
- Capability inventory: The skill possesses the capability to spawn background subagents, perform file system writes (state files, logs, backups), and potentially execute git operations via subagents.
- Sanitization: There is no evidence of sanitization or validation of the input items before they are dispatched for autonomous processing.
- [COMMAND_EXECUTION]: The skill dispatches work to subagents by spawning background processes. While this is core functionality, the combination with untrusted input (Category 8) increases the risk that an injected prompt could influence the commands executed by the autonomous subagents.
Audit Metadata