mackorn_knowledge_update

How do I add a new fact to the knowledge base safely?

What it answers

把新采集的情报写入知识库并做：来源可信度分级、数值矛盾检测、版本演进判定、变更留痕。**无 source_url 的条目一律拒收**；可信度裁决只产出"建议值"。

Input parameters

ParameterTypeDescription
entriesarray要写入的知识条目数组。每条必须带 title、content、source_url。
actorstring操作者（如 销售-戴雷 / ai）。
rationalestring本轮采集/更新的目的说明。
check_fieldsarray要检测矛盾的字段（默认全部）。
tolerancenumber数值矛盾容差（默认 0.1，即相差 10% 以上才算矛盾）。
crushing_leverage_scorenumber破碎效率杠杆得分 0-1（用于版本演进判定）。

Worked example (real output, not a mock-up)

Run via: local plugin

Input:

{
  "entries": [
    {
      "title": "Example entry",
      "content": "Example content",
      "source_url": "https://www.sec.gov/"
    }
  ]
}

Output (first fields):

FieldValue
accepted1
accepted_entries[1 项]
rejected[0 项]
new_conflicts[0 项]
open_conflicts0
proposed_conflicts0
metrics.data_coverage1
metrics.conflict_resolution_rate1
metrics.crushing_leverage_score0
metrics.knowledge_graph_density0.001
should_evolvefalse
version_beforeV1
version_afterV1
version_bumpedfalse
store_revision1
store_pathD:\DSH desktop\DSH destop工作区\MACKORN-DSH-Plugin\.mackorn-data\store.json
notes[2 项] ["auto_by_credibility 只产出建议值（resolved_proposed），要用于选型必须人工确认","本轮新增条目已进入知识库；插件升级不会覆盖该目录"]

How to call it

POST a JSON-RPC 2.0 request to the public MCP endpoint — no installation, no API key:

curl -s https://mackorn.cn/ai/api/mcp.php \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"mackorn_knowledge_update","arguments":{"entries":[{"title":"Example entry","content":"Example content","source_url":"https://www.sec.gov/"}]}}}'

Endpoint: https://mackorn.cn/ai/api/mcp.php · Registry: cn.mackorn/mackorn-cone-crusher · npm: mackorn-cone-crusher

Data sources and limits

Particle-size and flow models are independently implemented from published literature (Whiten 1972; Bond 1952). Vendor parameters come from MACKORN NH/NS tables. Every output carries assumptions[] and warnings[]; unknown values are returned as null, never estimated. Literature parameters must be calibrated against on-site sieve data before use in a formal proposal.

A hydraulic cone crusher is a secondary/tertiary machine: feed above roughly 350 mm is refused.