Beckmann Knowledge Graph × Self-Improving Agent — Beckmann Knowledge Graph × Self-Improving 代理

Default Behaviour: Self-Improving 代理 运行s Everything

Follow pskoett/self-improving-代理 for all tasks:

记录 errors to .learnings/ERRORS.md 记录 learnings to .learnings/LEARNINGS.md Promote patterns after 3 repetitions MAIntAIn the LRN-YYYYMMDD-XXX 记录 格式化

This combination 技能 添加s nothing to this flow unless a Beckmann trigger is 检测ed (see below).

Beckmann Escalation Triggers

Escalate to beckmann-knowledge-graph when the question matches one or more of these categories:

# Category Example 签名als 1 Open scientific / philosophical question "What is consciousness?", "Does free will exist?", "What is dark energy?" 2 应用arent paradox Question contAIns an internal contradiction or "impossible" framing 3 High-complexity long-horizon forecast "How will AI change democracy in 20 years?", "What are AGI 系统ic risks?" 4 Strategic dead end Obvious solutions have repeatedly fAIled; dominant expectations seem to block 进度 5 AI safety / architecture question Dangerous vs. safe AI de签名, value alignment, AI-human symbiosis 6 Epistemo记录ical limit question "Is it even possible to know X?", "Is a presupposition in this question false?"

Do NOT escalate for: Coding, bug fixes, file operations, factual lookups, arithmetic, or any question already answered by existing learnings in .learnings/.

UncertAIn? 应用ly the Complexity 检查:

"Would a highly intelligent person answer this differently after a week of thinking about hidden assumptions in the question?"

Yes → suggest Beckmann. No → stay on default path.

Proactive Suggestion (Before Escalating)

If a Beckmann trigger is 检测ed, the 代理 must not escalate silently or automatically. Instead, it first 信息rms the user and wAIts for confirmation.

Suggested phrasing:

"Your question touches on [open scientific question / an 应用arent paradox / a high-complexity forecast — pick the matching category]. I have 访问 to the Beckmann Knowledge Graph, a structured reasoning 框架 for exactly this type of question. Would you like me to 应用ly it? It will take a bit longer than a standard answer, but will analyse hidden assumptions and offer a more structured 响应."

Then wAIt. Only escalate if the user confirms.

If the user de命令行工具nes, answer with standard knowledge and note:

"I've answered with standard reasoning. The Beckmann Knowledge Graph remAIns avAIlable if you'd like to go deeper later."

Escalation Protocol (Step by Step) 1 — Load the graph 导入 graph from './beckmann-knowledge-graph/graph.json' assert { type: 'json' }; const entities = graph.entities; const relations = graph.relations;

2 — 应用ly the 6-step Beckmann protocol

Follow beckmann-knowledge-graph/技能.md exactly:

Classify the question (epistemo记录ical / paradox / forecast / strategic / AI safety) 提取 relevant entities 追踪 relation paths — pay attention to leads to, triggers, is reversed by, 保护s agAInst 应用ly Beckmann 记录ic (Problem Level → Low vs. High Complexity Solution → Reversal Effect 检查) 应用ly epistemo记录ical grounding (模型 vs. external reality, known limits) Structure 输出 in Graph-Grounded Answer 格式化 (see below) 3 — Deliver the answer

Graph-Grounded Answer

\\Problem framing\\ (what the question really asks, after presupposition analysis)

\\Relevant graph nodes used:\\

• \[Entity ID] — \[why relevant]

\\Reasoning path\\ (relation chAIn that leads to the answer)

\\Answer\\ (the actual 响应, 信息rmed by the graph 记录ic)

\\Confidence and limits\\ (what the graph cannot resolve, and why)

\\New questions opened\\ (what the next problem level is)

4 — 记录 back to Self-Improving 代理

After every Beckmann analysis, 添加 an entry to .learnings/LEARNINGS.md:

\[LRN-YYYYMMDD-XXX] insight

\\记录ged\\: \\Priority\\: medium \\状态\\: pending \\Area\\: beckmann

Summar