Unverified50% confidenceFactExact time
The tool is driven by three command-line arguments, --product, --specs and --local-engine, of which --specs supports an inline JSON string or a JSON file path.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedIn the tool-calling mechanism, developers attach a tool description list in JSON Schema format when calling the LLM API, and the model outputs call parameters in structured JSON format72% similarUnverifiedJSON Schema constrains the shape of JSON data through keywords such as type, properties, required, enum, and minimum, but in tool-calling scenarios it cannot express policy constraints such as "whether a path is allowed to be written to", which must be left to the permission and approval layer.71% similarUnverifiedHeadless mode runs detached from an interactive terminal via the claude -p command, and --output-format json can be specified to embed output into scripts, with JSON metadata containing information such as total_cost, num_turns, and duration_ms.68% similarUnverifiedLLM本身并不直接执行工具，而是生成一个JSON字符串指定要调用的工具名和参数，由Agent执行逻辑解析并实际调用对应函数68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1064609API
curl https://kongchang.com/api/v1/knowledge/claims/1064609MCP
get_claim(id=1064609)