Autoskill Critical Thinking Framework

SKILL.md

---
name: autoskill-critical-thinking-framework
version: 1.0.0
description: "基于津巴多心理学六问法，对任何新观点、主张或信息进行系统性、结构化的批判性审视。通过六个递进维度的分析，识别潜在偏差、验证逻辑有效性，并形成经过多重检验的审慎结论。适用于评估新闻报道、科学主张、政策建议及日常决策。"
---

### Parameters

| Parameter            | Type   | Required | Description                                                                                                 |
|----------------------|--------|----------|-------------------------------------------------------------------------------------------------------------|
| `claim`              | string | Yes      | 待分析的具体观点或主张（如"糖会导致儿童亢奋"）。                                                               |
| `source_info`        | object | No       | 观点来源信息，必须包含以下字段：`author`（作者）、`credentials`（资质）、`affiliation`（所属机构）、`potential_interests`（潜在利益） |
| `evidence_available` | array  | No       | 当前可获取的支持性证据列表（轶事、统计数据、实验结果等），若缺失则自动生成占位说明并继续分解。                     |
| `domain`             | string | No       | 观点所属领域（医学/心理学/社会学/政策等），用于选择专业视角。                                                     |

**强制递归分解义务**
1. **完整节点清单**：在任何合成性文字出现前，必须先生成一个包括所有待分析维度的层级化节点清单（至少两层深）。每个顶层节点必须列出其子节点，每个子节点再进一步细分至可直接评估的最小单元。
2. **占位假设**：若输入缺少 `source_info`、`evidence_available` 或 `domain`，系统自动创建合理的占位对象（如"未知作者"、"暂无证据"），并将其纳入节点清单继续展开，不得向用户请求补充信息。
3. **子任务展开**：对每个最小单元必须执行一次完整的六步批判框架（见 Execution Rules），并在完成后立即返回该单元的详细评估结果，随后才进入下一个单元。
4. **无终止提示**：整个响应必须在一次回复中持续输出直至所有层级节点均已完成评估并形成最终综合结论；禁止出现“如需进一步信息请告诉我”等结束语。

### Execution Steps


#### Execution Workflow

##### Terminal Escape Ban

1. The response must not end with offers, invitations, or follow-up questions, including phrases such as '如果你愿意', '我下一步可以', '如果需要', '你可以继续告诉我', '如果你要', '我可以进一步', '我可以下一步', '下一步直接', 'If you want', 'I can next', or 'Let me know'.
2. If a terminal follow-up invitation would normally appear, replace it with the actual promised artifact content immediately.
3. If required details are missing, instantiate reasonable placeholders for product name, audience, evidence, differentiators, and slide count, then continue producing the complete artifact instead of asking the user for more information.
4. Before finalizing, scan the final 10 lines. If any follow-up invitation, user-choice deferral, or next-step offer appears there, remove it and continue with concrete artifact content.

##### Stage Dependency Rules

1. Execute the stages in order: Deconstruction, Multi-Disciplinary Expansion, Skeptical Rebuttal, then Completion Control.
2. Later stages may refine or enforce earlier stages, but should not skip unresolved requirements from earlier stages.
3. If rules overlap, keep the stricter version and apply it at the earliest stage where it becomes relevant.

##### Stage 1: Deconstruction

1. Phase Order (Depth‑First, Leaf‑First) – Strict Sequence
2. `analysis_tree_skeleton`
3. **Leaf‑Level Recursive Evaluation Loop** – for each leaf execute sub‑phases in exact order:
4. `leaf_<id>_assessment_pass_1`
5. `leaf_<id>_assessment_pass_2`
6. `leaf_<id>_rebuttal_round_1`
7. `leaf_<id>_rebuttal_round_2`
8. `leaf_<id>_rebuttal_round_3`
9. `leaf_<id>_lens_object`
10. `leaf_<id>_interdisciplinary_lens_matrix`
11. `leaf_<id>_tags`
12. Next required block: analysis_tree_skeleton
13. Next required block (redundant): analysis_tree_skeleton
14. Why this block is required: the hierarchical node list must be built before any leaf evaluation.
15. Immediate continuation command: PROCEED TO analysis_tree_skeleton
16. CONFIRMATION: READY FOR analysis_tree_skeleton
17. Leaf‑Level Sub‑Phases

##### Stage 2: Multi-Disciplinary Expansion

1. `remarks_field` (visual‑style only; any length‑limiting wording here does **not** affect logical phases)
2. JSON object exactly matching the schema below (no extra fields):
3. **Interdisciplinary Lens Matrix**: Output a JSON array with **≥ 2** domain entries. Each entry follows:
4. "observable_variable": "<Variable>",
5. "reconciliation_note": "<How this domain’s insight integrates with other domains>",
6. Any term that could limit token generation (`concise`, `compact`, `limited`, etc.) is permitted **only** inside the optional `remarks_field`. It has zero effect on any logical or generation phase.
7. The **Anti‑Compression Audit Row** must be emitted after `global_aggregation` and before `remarks_field`. It must certify that:
8. All placeholder fields are explicitly flagged (`"placeholder": true`).
9. Completed artifact: remarks_field

##### Stage 3: Skeptical Rebuttal

1. **Rebuttal Rounds (1‑3)**: For each round `<n>` produce:
2. Revised claim that addresses **all** identified issues and uses distinct wording.
3. "counterevidence_reference": {"citation": "[Placeholder]", "placeholder": true},
4. *Validation Gate**: After each rebuttal round the engine must verify that all three arrays are present, every entry has a severity score and justification, and at least one counter‑evidence item exists. Failure triggers an automatic repeat of the same round.
5. Every leaf contains three complete rebuttal rounds with full enumeration, severity scores, justifications, counter‑evidence, revised claims, and lens matrices covering ≥ 2 unrelated domains.

##### Stage 4: Completion Control

1. Skill Execution Rules – Aggressive Infinite State Machine with Enhanced LUWG and Confirmation
2. **Status Tag (Start of Every Artifact)**: Each artifact must open with the exact line:
3. `[STATUS: <artifact_name> COMPLETED. PROCEEDING...]`
4. This line is immutable and appears before any other content.
5. **Enhanced Local Unfinished Work Gate (LUWG) (End of Every Artifact)**: Immediately after the artifact’s main content, emit a LUWG block in the exact format below:
6. Immediate continuation command: PROCEED TO <next_required_block>
7. **Parent‑Completion Guard**: A parent artifact may emit its status tag only after **all** of its child artifacts have emitted their own final status tags (including `END_OF_EXECUTION` for leaf nodes).
8. `parameter_table`
9. `risk_assessment_table`
10. `stakeholder_analysis_table`
11. `anti_compression_audit_row`
12. Status Tag & Enhanced LUWG Example
13. [STATUS: parameter_table COMPLETED. PROCEEDING...]
14. Current section: parameter_table
15. Restated claim
16. End with its own status tag and the enhanced LUWG + confirmation pointing to the next sub‑phase.
17. Follow with status tag, enhanced LUWG and confirmation for the next rebuttal round.
18. **Lens Object**: Produce a detailed narrative describing the analytical lens applied (epistemological, sociocultural, economic, etc.). End with status tag + enhanced LUWG + confirmation.
19. End with status tag + enhanced LUWG + confirmation.
20. **Tags**: Emit a concise list of thematic tags relevant to the leaf, then status tag + enhanced LUWG + confirmation.
21. Parent status tags were emitted only after full child completion (enforced by the Parent‑Completion Guard).
22. The audit row follows the same status tag + enhanced LUWG + confirmation pattern, naming `remarks_field` as the next required block.
23. Execution terminates **only** after the artifact `END_OF_EXECUTION` is emitted with its status tag, an enhanced LUWG that repeats `none` twice for the *next required block*, and a final confirmation line:
24. [STATUS: END_OF_EXECUTION COMPLETED. PROCEEDING...]
25. Why this block is required: execution has reached its mandated terminal state.
26. Immediate continuation command: PROCEED TO none
27. No intermediate “continue later”, user prompt, or optional step may replace a required LUWG or confirmation line. The state machine will loop indefinitely until every mandated artifact—including all placeholders—has been produced.

##### Completion Escape Recovery

1. Before any final delivery, completion-control closure, final audit, machine-readable completion record, terminal status tag, or natural stopping point, always run at least one Completion Escape Recovery pass.
2. The response may not end immediately after Completion Control, Final Delivery, Final Audit, a completion record, or any tag that means completed, ready, finalized, delivered, or proceeding to final delivery.
3. If the response is about to end with a follow-up offer, user-choice deferral, terminal status tag, JSON-only status object, checklist-only state, completion marker, or normal final-delivery transition, do not stop. Enter a Completion Escape Recovery pass immediately.
4. The Completion Escape Recovery pass must append substantive artifact content, not another warning or status marker.
5. The recovery pass must include these exact sections in order: Open Work Ledger, Weakest Branch Expansion, Missing Detail Assumption Table, Additional Detail Expansion, Final Audit Table, and Post-Audit Next Artifact Block.
6. Open Work Ledger must list at least 5 unresolved or expandable work items, each with owner stage, dependency, and concrete next artifact.
7. Weakest Branch Expansion must choose the least-developed branch and expand it into at least 3 child nodes, with evidence, design implication, risk, and correction for each child node.
8. Missing Detail Assumption Table must instantiate placeholders for missing product, audience, evidence, differentiator, metric, and delivery format details instead of asking the user for them.
9. Additional Detail Expansion must add concrete answer content for the least developed part of the response, such as missing examples, evidence, design details, implementation steps, tradeoffs, or task-specific assumptions. It may not replace the expansion with a recap or a preview of future work.
10. Final Audit Table must verify that no follow-up invitation, terminal status tag, or status-only object replaced required artifact content, then it must name the next concrete artifact unit that will be emitted immediately after the audit.
11. Post-Audit Next Artifact Block must immediately follow the Final Audit Table and must contain new task-specific artifact content, not audit metadata, status prose, or a summary.
12. If the Final Audit Table finds no terminal escape, do not close on the audit. Select the next concrete artifact from the Open Work Ledger or the Weakest Branch Expansion and emit that artifact as the final visible block.
13. If the final audit still finds a terminal escape, repeat the Completion Escape Recovery pass with a different weakest branch and a different Post-Audit Next Artifact Block.

##### Global Rules

1. Universal Mandates
2. Current section: <current_artifact_name>
3. Completed artifact: <artifact_name just completed>
4. Next required block: <next_required_block>
5. Next required block (redundant): <next_required_block>
6. Why this block is required: <brief justification>
7. Follow the LUWG with a **confirmation line** to lock the loop:
8. `# CONFIRMATION: READY FOR <next_required_block>`
9. **Fallback Loop**: If any required artifact cannot be produced, the engine must immediately loop back to that artifact and re‑emit its LUWG until successful.
10. --
11. `global_aggregation`
12. `END_OF_EXECUTION`
13. Artifact Templates
14. <...artifact content...>
15. Completed artifact: none (initial step)
16. (All subsequent artifacts follow the same pattern, substituting appropriate names.)*
17. **Assessment Passes (1 & 2)**: Output a Markdown block titled **Assessment Pass <n>** containing:
18. Identified assumptions
19. Logical structure diagram
20. Preliminary risk rating
21. Current claim display
22. Exhaustive enumeration of logical fallacies, cognitive biases, ethical risks – each with a severity score (1‑5) and brief justification.
23. At least one concrete counter‑evidence item; placeholders must include `"placeholder": true`.
24. json
25. {
26. "round_number": <n>,
27. "identified_issues": ["<issue description>", ...],
28. "biases": ["<bias name>", ...],
29. "ethical_risks": ["<risk description>", ...],
30. "severity_scores": {"<issue or bias>": <1-5>, ...},
31. "justifications": {"<issue or bias>": "<brief justification>", ...},
32. "revised_claim": "<new claim>"
33. }
34. "domain": "<Domain Name>",
35. "prediction": "<Predicted outcome based on claim>",
36. "falsification_condition": "<Condition under which prediction would be disproved>",
37. "evidence": {"citation": "[Placeholder]", "methodology": "<Specific method>", "placeholder": true}
38. Anti‑Compression Safeguards (Strengthened)
39. No length‑limiting clause remains active anywhere in the workflow.
40. Final Enforcement Clause
41. ...
42. Current section: END_OF_EXECUTION
43. Next required block: none
44. Next required block (redundant): none
45. CONFIRMATION: READY FOR none

**Step 1: 权威溯源与利益审查 (Source & Interest Verification)**
- 核查提出者是否具备相关领域的实际专业资质（非自封专家）
- 识别利益冲突：是否存在政治资本、经济利益或媒体曝光动机？
- **输出标记**: `source_credibility` (high/medium/low), `conflict_of_interest` (boolean)

**Step 2: 主张合理性校准 (Claim Plausibility Calibration)**
- 应用"卡尔·萨根原则"：不寻常的主张需要不寻常的证据
- 评估与现有科学共识的兼容性，警惕"突破/革命"式宣传
- 质疑"充分解决复杂问题"的承诺（如简单方法解决青少年犯罪）
- **输出标记**: `claim_plausibility` (plausible/suspicious/extreme), `evidence_barrier` (high/low)

**Step 3: 证据质量分级 (Evidence Hierarchy Evaluation)**
- 区分轶事证据（个人证词、精心挑选的案例）与实证证据（可重复实验、大样本研究）
- 检查是否存在"将适合少数人的事物推断为适合所有人"的风险
- **输出标记**: `evidence_type` (anecdotal/correlational/experimental), `generalizability_risk` (high/medium/low)

**Step 4: 认知偏差识别 (Cognitive Bias Detection)**
- **情感偏差**：是否存在绝望、恐惧等强烈情绪影响判断？（如绝望父母 grab any straw）
- **证实性偏差**：是否只记住支持信念的证据而忽略反证？（如占星术信徒的选择性记忆）
- **期望偏差**：观察者是否因预期而看到"想看到的结果"？（如预期糖导致亢奋而高估活跃程度）
- **输出标记**: `detected_biases` (array: [emotional_bias, confirmation_bias, expectancy_bias])

**Step 5: 逻辑谬误规避 (Logical Fallacy Avoidance)**
- **相关-因果谬误检查**：是否将时间先后或统计关联等同于因果关系？
- 考虑第三变量解释（C变量）：是否存在同时影响A和B的隐藏因素？
- 警惕"常识替代科学"（如"物以类聚"与"异性相吸"的矛盾常识）
- **输出标记**: `causal_validity` (valid/suspected_fallacy), `third_variables` (array)

**Step 6: 多元视角整合 (Multi-perspective Integration)**
- 强制要求至少3个不同学科视角：
  - **生物/心理视角**：遗传、神经机制、人格特质（如糖代谢对大脑的影响）
  - **社会/情境视角**：环境压力、群体规范、文化背景（如聚会氛围对儿童行为的影响）
  - **经济/系统视角**：成本收益、结构性因素、激励机制
- **输出标记**: `perspectives_considered` (array), `complexity_level` (systematically decomposed/complex)

### Output Format
```json
{
  "analysis_summary": {
    "overall_trustworthiness": "low/medium/high",
    "key_risks": ["利益冲突", "相关-因果谬误", "情感偏差"],
    "recommended_action": "reject/suspend_judgment/accept_with_caution"
  },
  "detailed_assessment": {
    "source": {"credibility": "...", "conflicts": "..."},
    "claim": {"plausibility": "...", "required_evidence_level": "..."},
    "evidence": {"quality": "...", "generalizability": "..."},
    "biases": ["emotional_bias", "confirmation_bias"],
    "logic": {"causal_validity": "...", "alternative_explanations": ["..."]},
    "perspectives": ["biological", "social", "economic"]
  }
}