jianying-harness
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill interacts with the local system using a Rust-based CLI tool (
jianying) to manage video editing jobs, audit tasks, and serve as a Model Context Protocol (MCP) transport. The instructions strictly limit execution to verified capabilities and require precise approval bindings for any command that modifies the system or consumes resources. - [INDIRECT_PROMPT_INJECTION]: The skill processes external media metadata, draft files, and JSON-based job configurations. This ingestion surface is mitigated by comprehensive validation steps defined in
Step 3(input probing), schema enforcement (jianying-job/v2), and a requirement for post-execution structural verification to ensure output integrity. - [DATA_EXPOSURE]: There are explicit rules prohibiting the collection, recording, or uploading of unrelated user credentials. The skill mandates the use of isolated target paths for all write operations, ensuring that original source drafts are never overwritten or exposed to unintended modifications.
Audit Metadata