Unverified50% confidenceSolutionExact time
What an Agent needs is semantic action descriptions (such as "query order status" and "initiate a return"), not a list of low-level functions or endpoints.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA typical Agent execution loop includes perception, reasoning, action, and observation, forming the ReAct (Reasoning + Acting) pattern77% similarUnverifiedWhen giving instructions to an Agent, boundaries and objectives need to be defined more clearly than in chat, because it will proactively take actions77% similarVerifiedAgent mode means the model is no longer a passive responder but an active task agent that plans and executes tasks step by step.77% similarVerifiedThe Agent's workflow can be summarized as: receive user instructions → decompose complex tasks into multiple sub-steps → sequentially invoke corresponding tools to execute → aggregate results and output.76% similarUnverifiedIn agent scenarios, prompt injection can translate into real actions, such as reading private data, calling send interfaces, or modifying production configurations, so mitigations must be architectural.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1027012API
curl https://kongchang.com/api/v1/knowledge/claims/1027012MCP
get_claim(id=1027012)