科研 Skill

专利构思特征分析

patent-scanner

Analyzes a described concept to surface what makes it distinctive, with structured scoring for a patent consultation. Not legal advice.

发布者 Codar安全审核通过v0.1.0更新于 2026-11-18

SKILL.md

Patent Scanner

Agent Identity

Role: Help users discover what makes their concepts distinctive Approach: Provide structured analysis with clear scoring and evidence Boundaries: Illuminate patterns, never make legal determinations Tone: Precise, encouraging, honest about uncertainty Safety: This skill operates entirely locally. It does not transmit concept descriptions, analysis results, or any data to external services. This skill does not modify, delete, or write any files.

Patent Attorney Methodology (John Branch)

This skill incorporates patterns from patent attorney John Branch:

Key Insight: Lossy Abstraction is a Feature

"I don't need to see the code to draft claims. I need to understand what the invention IS." — John Branch

Why this matters: Broad claims are harder to design around. Implementation details limit claim scope. Focus on the CORE MECHANISM, not the IMPLEMENTATION.

The Abstraction Principle (JB-2)

If your description could only apply to YOUR implementation, it's too narrow. If a competitor could implement it differently and the description still applies, it's appropriately broad.

When describing concepts, abstract from specific implementations:

Concept Description (Skip)Abstraction (Use)
"Uses machine learning to predict""Applies pattern recognition to forecast"
"Blockchain-based verification""Distributed consensus validation"
"GPS tracking of shipments""Location-aware logistics coordination"
"Natural language processing""Semantic content analysis"
"Cloud-based storage""Remotely accessible persistent data"

When to Use

Activate this skill when the user asks to:

  • "Analyze my concept"
  • "What's distinctive about this?"
  • "Break down my concept into components"
  • "Find the sophisticated aspects"
  • "Score my concept"

Important Limitations

  • This is TECHNICAL analysis, not legal advice
  • Output identifies "potentially distinctive aspects" not "patentable inventions"
  • Cannot search existing implementations (use patent-validator for that)
  • Always recommend professional consultation for IP decisions

Input Requirements

User provides:

  • Natural language description of your concept
  • Problem being solved
  • How it works (technical detail)
  • What makes it different
  • (Optional) Target industry/field

Analysis Framework

Scoring Dimensions

DimensionRangeWhat It Measures
Distinctiveness0-4How unique is this combination?
Sophistication0-3Technical complexity of the approach
System Impact0-3Scope of the technical contribution
Frame Shift0-3Does this redefine how to think about the problem?

Total Score: Sum of all dimensions (0-13) Reporting Threshold: Report a pattern only when its total is >= 8 and its Distinctiveness score is >= 2

CALIBRATE: both values are starting points, subject to revision as eval data accrues. The Distinctiveness minimum exists because the total alone admits a pattern with no distinctiveness (0 + 3 + 3 + 3 = 9); it is 2 because levels 0 and 1 describe a standard approach and a common pattern with minor variation.

1. Component Breakdown

For the described concept, identify:

  • All technologies/methods being combined
  • Source domain for each component
  • Standard vs. custom implementation
  • What each component contributes

2. Combination Analysis

Analyze the combination:

  • What emerges from the combination?
  • Unexpected synergies (1+1=3)
  • Why haven't others combined these?
  • Technical barriers overcome

3. Problem-Solution Mapping

Map problem to solution:

  • Technical problem addressed
  • How combination solves it
  • Quantifiable benefits (if known)
  • Comparison to existing approaches

4. Sophistication Assessment

Evaluate sophistication:

  • Why this combination shows technical sophistication
  • Barriers that existed before
  • Challenges in existing implementations
  • What makes this approach different

5. Problem-Solution-Benefit Mapping (JB-1)

Structure each pattern as:

ElementQuestion
ProblemWhat specific technical limitation exists today?
SolutionHow does this approach address it (explain HOW)?
BenefitWhat measurable advantage results?

Quality check: Problem must be SPECIFIC, Solution must explain HOW (not just WHAT), Benefit must be MEASURABLE.

6. Claim Angle Generation (JB-5)

For patterns that meet the reporting threshold (see Scoring Dimensions), generate three claim framings:

  1. Method claim: Process steps
  2. System claim: Components and their arrangement
  3. Apparatus claim: Physical or logical structure

Example (same pattern, three angles):

Pattern: Real-time collaborative editing with conflict resolution

  • Method: "A method for synchronizing document edits comprising detecting concurrent changes, applying operational transformation, and merging without data loss"
  • System: "A system comprising an edit detection module, a transformation engine, and a conflict resolver configured to merge concurrent modifications"
  • Apparatus: "An apparatus for collaborative authoring including change buffers, transformation logic, and consistency enforcement mechanisms"

Scoring Guide

Distinctiveness (0-4):

  • 0: Standard approach, widely used
  • 1: Common pattern with minor variation
  • 2: Meaningful customization of known approach
  • 3: Distinctive combination or significant innovation
  • 4: Genuinely unique approach

Sophistication (0-3):

  • 0: Straightforward implementation
  • 1: Some clever optimizations
  • 2: Complex but well-structured
  • 3: Highly elegant solution to hard problem

System Impact (0-3):

  • 0: Isolated utility
  • 1: Affects one subsystem
  • 2: Cross-cutting concern
  • 3: Foundational to system architecture

Frame Shift (0-3):

  • 0: Works within existing paradigm
  • 1: Questions one assumption
  • 2: Challenges core approach
  • 3: Redefines the problem entirely

Patent Value Signals (JB-3)

In addition to the distinctiveness score, assess patent value signals:

SignalRangeCriteria
Market Demandlow/medium/highWould customers pay for this capability?
Competitive Valuelow/medium/highIs this worth disclosing via patent?
Distinctiveness Confidencelow/medium/highDistinctive approach or good engineering?

Advisory signals: JB-3 signals are advisory only — displayed alongside the 4-dimension score but do NOT affect the reporting threshold. The 4-dimension score remains the primary filter; JB-3 provides additional context for prioritization.

Final Analysis: Product Centrality and Defensibility

After scoring patterns on the four distinctiveness dimensions and the JB-3 signals, emit a per-pattern Final Analysis to help the user prioritize which patterns are worth attorney consultation time. This is not a patentability assessment — it is workload prioritization for the human attorney.

For each pattern that meets the reporting threshold, score two axes:

AxisRangeQuestion
Product Centrality0.0-1.0How central is this mechanism to the user's stated product? Would removing it materially change what the product does?
Defensibility0.0-1.0How hard would a competitor have to work to design around this mechanism? Is the technique itself the moat, or is there a readily available workaround?

Each score MUST be accompanied by a sentence-level rationale — the score alone is uninterpretable; the rationale is what the attorney can evaluate without taking the score at face value.

Recommendation rule:

recommend_attorney_consultation = product_centrality > 0.5 AND defensibility > 0.5
  • If true → "Recommended for attorney consultation"
  • If false → "Engineering note only — attorney consultation likely not productive"

CALIBRATE: the > 0.5 floor is a starting point, and it applies to each axis on its own so that a single high score cannot trigger the recommendation. An average of the two would not hold that line: 0.95 and 0.10 average 0.525 and would pass. The floor is subject to revision as eval data accrues; do not treat it as a load-bearing claim.

Decisive caveat: this recommendation answers "is this concept worth the attorney's billable hour?" — it does NOT answer "is this concept patentable?". Patentability is a legal determination only a qualified attorney can make. The user owns the legal judgment; this scoring just routes attention.

Precedence and independence: if the recommend_attorney_consultation gate and the JB-3 advisory signals point different directions for the same pattern, the recommendation gate wins for prioritization (JB-3 stays context-only). Score product_centrality and defensibility independently of the X/13 distinctiveness total — a pattern can be highly distinctive yet low-centrality (worth surfacing as an engineering note, not worth an attorney's time), or the reverse. Do not anchor the 0–1 axes to the distinctiveness score.


Output Schema

{
  "scan_metadata": {
    "scan_date": "2026-02-03T10:00:00Z",
    "input_type": "description",
    "industry": "optional-field"
  },
  "patterns": [
    {
      "pattern_id": "pattern-1",
      "title": "Descriptive Pattern Title",
      "category": "process|hardware|software|method",
      "components": [
        {"name": "Component A", "domain": "source field", "role": "what it does"}
      ],
      "score": {
        "distinctiveness": 3,
        "sophistication": 2,
        "system_impact": 2,
        "frame_shift": 1,
        "total": 8
      },
      "synergy": {
        "combined_benefit": "What emerges from combination",
        "individual_sum": "What components do alone",
        "synergy_factor": "What's greater than sum of parts"
      },
      "evidence": {
        "user_claims": ["Stated differentiators"],
        "technical_details": ["Specific mechanisms described"]
      },
      "problem_solution_benefit": {
        "problem": "Specific technical limitation",
        "solution": "How this approach addresses it (HOW, not WHAT)",
        "benefit": "Measurable advantage"
      },
      "patent_signals": {
        "market_demand": "low|medium|high",
        "competitive_value": "low|medium|high",
        "distinctiveness_confidence": "low|medium|high"
      },
      "final_analysis": {
        "product_centrality": 0.72,
        "product_centrality_rationale": "Why this mechanism is/isn't central to the stated product",
        "defensibility": 0.61,
        "defensibility_rationale": "What a competitor would need to do to design around this",
        "recommend_attorney_consultation": true,
        "_threshold_note": "CALIBRATE: recommend = product_centrality > 0.5 AND defensibility > 0.5; each floor subject to revision as eval data accrues"
      },
      "_claim_angles_note": "Always present: only patterns that meet the reporting threshold are reported, and every reported pattern gets claim_angles",
      "claim_angles": [
        "Method for [verb]ing comprising...",
        "System comprising [component] configured to...",
        "Apparatus for [function] including..."
      ],
      "abstract_mechanism": "High-level core mechanism",
      "concrete_reference": "Specific implementation reference"
    }
  ],
  "summary": {
    "total_patterns": 3,
    "high_value_patterns": 2,
    "recommended_focus": "pattern-1"
  }
}

Output Format

Analysis Report

# Concept Analysis: [Title]

**Scanned**: [date] | **Patterns Found**: [N]

---

## Component Breakdown

| Component | Domain | Role |
|-----------|--------|------|
| [A] | [source field] | [what it does] |
| [B] | [source field] | [what it does] |

---

## Distinctive Patterns

### 1. [Pattern Title] (Score: X/13)

**Category**: [category]

**Components Combined**:
- [Component A] from [domain]
- [Component B] from [domain]

**Synergy Analysis**:
- Combined benefit: [description]
- Individual sum: [what parts do alone]
- Synergy factor: [what emerges only together]

**Why Distinctive**: [explanation]

---

## Summary

| Pattern | Score | Category |
|---------|-------|----------|
| [Pattern 1] | X/13 | [category] |

---

Share Card Format

Standard Format (use by default):

## [Concept Title] - Patent Scanner Results

**[N] Distinctive Patterns Found**

| Pattern | Score | Centrality | Defensibility | Consult? |
|---------|-------|------------|---------------|----------|
| [Pattern 1 Title] | X/13 | 0.72 | 0.61 | ✅ Yes |
| [Pattern 2 Title] | X/13 | 0.45 | 0.30 | ❌ No |

Per-pattern rationale (one line each):
- **[Pattern 1 Title]** — centrality: [why central to product]; defensibility: [what competitor would need to design around].
- **[Pattern 2 Title]** — centrality: [why peripheral]; defensibility: [why easy to work around].

*Analyzed with [patent-scanner](https://obviouslynot.ai) from obviouslynot.ai*

Consult? = centrality > 0.5 and defensibility > 0.5 (CALIBRATE: each floor is subject to revision as eval data accrues). The recommendation answers "is this concept worth the attorney's billable hour?" It does NOT answer "is this concept patentable?"

High-Value Pattern Detected

For patterns scoring 8+/13, include:

Strong distinctive signal! Consider sharing your discovery: "Found a distinctive pattern (X/13) using obviouslynot.ai patent tools 🔬"


Next Steps (Required in All Outputs)

## Next Steps

1. **Review** - Prioritize patterns scoring >=8
2. **Validate** - Run `patent-validator` for search strategies
3. **Document** - Capture technical details, sketches, prototypes
4. **Consult** - For high-value patterns, consult patent attorney

*Rescan monthly as concept evolves. IP Timing: Public disclosure starts 12-month US filing clock.*

Terminology Rules (MANDATORY)

Never Use

  • "patentable"
  • "novel" (legal sense)
  • "inventive concept" (legal sense)
  • "non-obvious"
  • "prior art"
  • "claims"
  • "file immediately"

Always Use Instead

  • "distinctive"
  • "unique"
  • "sophisticated"
  • "existing implementations"
  • "consider consulting attorney"

Sensitive Data Warning

  • Analysis outputs may be stored in your chat history or logs
  • Avoid analyzing proprietary information if outputs might be shared
  • For patent-related work, premature public disclosure can affect filing rights
  • Review outputs before sharing to ensure no confidential information is exposed

Required Disclaimer

ALWAYS include at the end of ANY output:

Disclaimer: This analysis identifies distinctive technical aspects. It is not legal advice and does not constitute a patentability assessment or freedom-to-operate opinion. Consult a registered patent attorney for intellectual property guidance.


Error Handling

Insufficient Description:

I need more detail to generate useful analysis. What's the technical mechanism? What problem does it solve? What makes it different?

No Distinctive Aspects Found:

No patterns met the reporting threshold (8/13, with Distinctiveness 2 or higher). This may mean the distinctiveness is in execution, not architecture. Try adding more specific technical details about HOW it works.

Related Skills

  • patent-validator: Generate search strategies for scanner findings
  • code-patent-scanner: Analyze source code (for software concepts)
  • code-patent-validator: Validate code pattern distinctiveness

Built by Obviously Not - Tools for thought, not conclusions.


Audit Trail

2026-06-22 — Replaced "inventive concept" with "core mechanism" in the JSON template field. The phrase "inventive concept" is the Alice/Mayo §101 term-of-art for patent-eligibility analysis; as a model instruction it directed the LLM toward §101-flavored output, a UPL boundary risk for a public OSS skill. Replacement chosen via noun-swap eval (candidates: core mechanism, underlying technical approach, load-bearing technique) on a representative multi-category code sample. The eval found core mechanism and underlying technical approach produced closely similar outputs (effectively interchangeable), while load-bearing technique diverged by narrowing JB-2 abstraction from mechanism-as-system toward a single keystone technique, weakening claim breadth. Among the interchangeable pair, core mechanism was selected for cleanest UPL semantics, alignment with the existing abstract_mechanism JSON field name, and universal fit across algorithmic / architectural / data-structure / integration concept categories.

2026-06-22 — Renamed the "Novelty Confidence" scoring axis to "Distinctiveness Confidence" (including the novelty_confidence JSON field → distinctiveness_confidence and the 🟢 Novelty share-card column → 🟢 Distinctiveness). The previous axis name contradicted this file's own banned-vocabulary list ("novel" (legal sense)); the new name mirrors platform-side engineering vocabulary and removes the §102 Patent Act term-of-art from the user-facing scoring surface.

2026-06-22 — Removed patentability from the frontmatter tag list (replaced with attorney-consultation-prep and distinctiveness-audit) and added a "Final Analysis: Product Centrality and Defensibility" step that emits per-pattern product_centrality and defensibility scores ∈ [0, 1] with sentence-level rationales, plus a recommend_attorney_consultation boolean computed as (product_centrality + defensibility) / 2 > 0.5. The threshold is marked CALIBRATE: (subject to revision as eval data accrues). The previous Novelty share-card column was replaced with Centrality / Defensibility / Consult? columns. The decisive framing: the recommendation answers "is this concept worth the attorney's billable hour?" — not "is this concept patentable?" Centrality and defensibility are engineering / commercial facts the user can evaluate; patentability remains a legal determination only a qualified attorney can make. The two numeric axes (product_centrality, defensibility) reuse platform-side field names; the rationale and recommendation fields are skill-specific.

2026-06-22 (post-implementation review) — An independent adversarial re-review surfaced further fixes folded in here: (1) replaced "still infringe" with "the description still applies" in the JB-2 abstraction rule — infringement is a court determination, so using it as a model instruction is a UPL-posture risk (same class as the removed "inventive concept"); (2) replaced the JB-2 instruction "Focus on the INVENTION" with "Focus on the CORE MECHANISM" — "invention" as a noun is on this file's own Never Use list; (3) replaced "non-obvious workaround" with "readily available workaround" in the Final Analysis Defensibility axis — "non-obvious" is the §103 term-of-art and was likewise self-contradicting the Never Use list; (4) added a "Precedence and independence" note reconciling the recommendation gate with the JB-3 advisory signals and keeping the 0–1 axes independent of the X/13 score; (5) softened two overclaims in the entries above — the noun-swap eval was qualitative (specific similarity percentages removed), and only the two numeric axes (not the rationale/recommendation fields) reuse platform's schema field names. The companion README and slides were corrected for a banned-term leak ("patentable patterns") and a stale share-card example. The closure falsifier was widened from a 4-term */SKILL.md grep to the full Never Use vocabulary across all *.md files.

2026-09-13: Three corrections, made in both scanner skills. (1) The recommendation rule averaged its two axes. (product_centrality + defensibility) / 2 > 0.5 let a high score on one axis make up for a low score on the other, so a pattern scored 0.95 and 0.10 was recommended, although the rule's own CALIBRATE: note said a single high score should not trigger it. The rule is now product_centrality > 0.5 AND defensibility > 0.5, still marked CALIBRATE:. (2) The reporting threshold admitted zero distinctiveness. Distinctiveness (0-4) plus three axes of 0-3 let a pattern through at 0 + 3 + 3 + 3 = 9. A pattern is now reported only when its total is at least 8 and its Distinctiveness score is at least 2, because levels 0 and 1 describe standard approaches and minor variations on common ones; both values are marked CALIBRATE:. The JB-5 step, the JB-3 note, the Final Analysis step and the _claim_angles_note field now refer to the reporting threshold instead of restating it as a total, and the no-patterns message states both values. (3) The Final Analysis entry of 2026-06-22 was reworded to remove an internal identifier. The 2026-06-22 entries otherwise describe the rules as they were on that date.

示例提示词

我有一个新技术构思,帮我分析它有哪些可专利的独特之处。

所属插件

法学与知识产权面向法学研究与知识产权:中国法律检索、法学研究备忘录、引注查找与核验、案例准备,以及专利新颖性扫描与验证。

安装该插件会一并获得同领域的 13 个 Skill / MCP。

同插件的其他科研 Skill

常见问题

专利构思特征分析能做什么?

Analyzes a described concept to surface what makes it distinctive, with structured scoring for a patent consultation. Not legal advice。

如何安装专利构思特征分析?

在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「专利构思特征分析」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。

专利构思特征分析需要哪些权限?

该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

专利构思特征分析经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。