Unverified50% confidenceFactExact time
Even with the LLM temperature set to 0, the same intent may still generate different SQL structures due to differences in prompt wording, context length, and prior tool call results.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedSQL injection can fundamentally isolate data from instructions through parameterized queries, whereas there is currently no equivalent "parameterized" solution for the LLM inference process.62% similarUnverifiedPrompt injection shares a structural analogy with SQL injection vulnerabilities: the latter exploits a database's inability to distinguish SQL code from data strings, while the former exploits an LLM's inability to distinguish meta-instructions from the content being processed.62% similarUnverifiedAdding support for a new model at the SDK level typically requires registering a new model ID string and handling parameter allowlist differences (such as certain models not supporting temperature or top_p) as well as capability declarations.62% similarUnverified直接让 LLM 生成 SQL 存在语义漂移和幻觉问题，可能引用不存在的字段或表名60% similarUnverifiedData ontology improves the groundedness of model-generated SQL and reduces the probability of hallucination-type erroneous outputs by providing canonical aliases, definitional descriptions, and calculation logic for fields.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1045760API
curl https://kongchang.com/api/v1/knowledge/claims/1045760MCP
get_claim(id=1045760)