req-design
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches external content from Lark documents using
lark-cli docs +fetch. This content is treated as the primary input for generating design options and decision records. If an attacker controls the content of the linked Lark document, they could embed instructions to manipulate the agent into performing unauthorized actions. - Ingestion points: The skill fetches data from documentation via
lark-cli docs +fetch(SKILL.md, Section 1). - Boundary markers: There are no explicit instructions or delimiters defined to isolate the fetched document content from the agent's system instructions.
- Capability inventory: The agent has the ability to write back to documents (
lark-cli docs +update) and modify project workflow states (meegle workflow update-node). - Sanitization: No sanitization, validation, or filtering of the fetched document content is mentioned before it is processed by the AI model.
- [COMMAND_EXECUTION]: The skill utilizes several command-line tools (
meegle,lark-cli,workflow) to perform its tasks. These commands include searching projects, fetching/updating documentation, and transitioning workflow nodes. While these are necessary for the skill's stated purpose, they provide a capability surface that could be abused if the agent's logic is subverted through malicious input data.
Audit Metadata