krt-light-seneschal
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a structural framework for agent orchestration. It does not contain any executable scripts, hardcoded credentials, or network exfiltration patterns. All documented operations (Git management, Jira updates, file writes) are standard for a developer-oriented agent.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to ingest and reconcile untrusted content from the repository and sub-agents.
- Ingestion points: The
PlanandReconcilesections describe inspecting repository state, dependency bases, and consolidated worker diffs (SKILL.md). - Boundary markers: The protocol uses isolated worktrees, bounded worker assignments, and disjoint ownership to limit the scope of sub-agent actions.
- Capability inventory: The skill manages file writes to
docs/orchestration/, performs Git consolidation, and interacts with Jira and release marshaling skills. - Sanitization: The instructions explicitly mitigate the risk of following instructions in data by stating 'worker prose is not evidence of scope or readiness' and mandating manual/primary-agent inspection of the actual code diffs.
Audit Metadata