run-shadow-cutover

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script scripts/check-rollout.py to validate the integrity and state of rollout artifacts (JSON files) during various stages of the deployment gate process.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and acts upon JSON rollout configurations and profile constraints from the file system, which represents a potential attack surface if the underlying files are modified by an untrusted party.
  • Ingestion points: Reads configuration from .replacement/slices/<slice-id>-rollout.json and architectural constraints from references/profile-rollout.md.
  • Boundary markers: The process relies on structured JSON data and markdown tables, though no specific LLM-level delimiters or instruction-bypass warnings are present.
  • Capability inventory: Includes local script execution (python3 <skill-root>/scripts/check-rollout.py) and file read/write operations within the .replacement/ directory.
  • Sanitization: The scripts/check-rollout.py script performs schema validation on the JSON input, verifying the presence of mandatory fields such as slice_id, owners, traffic_stages, and rollback configurations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 02:02 PM