lark-shared
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes output from the lark-cli tool, which may include untrusted content from external documents or messages stored in the Lark/Feishu environment. * Ingestion points: CLI stdout and stderr output (JSON envelopes) as described in lark-shared-output-contract.md. * Boundary markers: The skill defines a structured JSON contract (ok, data, error) to delimit tool responses and guide parsing. * Capability inventory: Subprocess execution of lark-cli for file operations, authentication management, and resource access. * Sanitization: The instructions focus on structured parsing of JSON fields rather than explicit sanitization of string content.
- [SAFE]: The skill implements explicit security controls that mitigate common risks. It requires mandatory human confirmation for high-risk write operations (identified by exit code 10 and risk: high-risk-write), and it enforces a restriction to relative file paths for all file-related parameters to prevent path traversal attacks.
Audit Metadata