菜单
Decode BS — run text through structural, content, and rhetorical analysis to produce a FuzzyEnum<BsLevel> score with tetravalent mapping
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | demerzel-bs-decode |
| description | Decode BS — run text through structural, content, and rhetorical analysis to produce a FuzzyEnum<BsLevel> score with tetravalent mapping |
Analyze text for bullshit density using the BS detection grammar (grammars/gov-bs-generators.ebnf). Produces a quantified BS score, per-test breakdowns, rhetorical device inventory, tetravalent mapping, and a clear-speech rewrite when warranted.
/demerzel bs-decode [text or URL]
/demerzel bs-decode --file [path] — decode a file
/demerzel bs-decode --url [url] — fetch and decode a URL
/demerzel bs-decode --self [artifact-path] — self-check a Demerzel governance artifact
INPUT
│
▼
STRUCTURAL ANALYSIS
├─ Specificity Test → FuzzyEnum<Low|Medium|High>
├─ Falsifiability Test → FuzzyEnum<Zero|Low|Medium|High>
└─ Commitment Test → FuzzyEnum<Low|Medium|High>
│
▼
CONTENT ANALYSIS
├─ Density Test → FuzzyEnum<Fluff|Mixed|Substance>
├─ Factual Density → FuzzyEnum<None|Sparse|Dense>
└─ Actionability Test → FuzzyEnum<None|Vague|Actionable>
│
▼
RHETORICAL ANALYSIS
├─ Device Scan → List of devices from gov-bs-generators.ebnf §11
└─ Evasion Scan → List of tactics from gov-bs-generators.ebnf §12
│
▼
SCORING
│ Weighted combination of all tests
│ Weights: specificity 0.20, falsifiability 0.15, commitment 0.20,
│ content_density 0.15, factual_density 0.15, actionability 0.15
│
▼
CLASSIFICATION → FuzzyEnum<BsLevel>
│ Clear (score > 0.8) — real content
│ Mild (score 0.6-0.8) — some fluff, substance present
│ Moderate (score 0.4-0.6) — more style than substance
│ Serious (score 0.2-0.4) — almost entirely empty rhetoric
│ Pure (score < 0.2) — could be randomly generated
│
▼
TETRAVALENT MAPPING
│ Clear/Mild → T (verified real content)
│ Moderate → U (substance unclear, investigate)
│ Serious/Pure → C (contradictory — sounds meaningful, says nothing)
│
▼
REWRITE (if score ≤ Moderate)
│ Generate clear-speech alternative using antidote pattern:
│ claim + evidence + specifics + commitment
│
▼
OUTPUT
Replace the subject with any other subject. If the sentence still makes sense, specificity is LOW.
| Result | Criterion |
|---|---|
| High (1.0) | Statement only applies to the specific subject named |
| Medium (0.5) | Partially specific — some generic, some concrete |
| Low (0.0) | Could apply to any company, product, or person |
Ask: "What evidence would DISPROVE this claim?" If no evidence could disprove it, falsifiability is ZERO.
| Result | Criterion |
|---|---|
| High (1.0) | Clear criteria for disproof exist |
| Medium (0.5) | Partially testable — some aspects verifiable |
| Low (0.2) | Very difficult to test |
| Zero (0.0) | Unfalsifiable — no evidence could disprove it |
Ask: "WHO will do WHAT by WHEN?" Missing any element reduces the score.
| Result | Criterion |
|---|---|
| High (1.0) | All three present: who, what, when |
| Medium (0.5) | Two of three present |
| Low (0.0) | Vague aspiration with no specifics |
Remove all adjectives, adverbs, and buzzwords. Ratio of remaining content words to total words.
Count verifiable facts per sentence: numbers, dates, names, measurements, citations.
Could someone act on this statement without asking clarifying questions?
Scan for devices defined in grammars/gov-bs-generators.ebnf:
Devices (§11): weasel_words, false_precision, appeal_to_authority, false_dichotomy, circular_reasoning, loaded_language, appeal_to_novelty, appeal_to_tradition, bandwagon, thought_terminating_cliche, strategic_ambiguity
Evasion Tactics (§12): deflection, flooding, reframing, credentialism, complexity_shield, future_promise, whataboutism
Report: devices per sentence (rhetorical density) and specific instances found.
Match input against BS domain productions in the grammar:
| Domain | Grammar Section |
|---|---|
| Consulting | §1: consulting_bs |
| AI/Tech | §2: ai_bs |
| Academic | §3: academic_bs |
| Motivational | §4: motivational_bs |
| Governance | §5: governance_bs |
| Corporate | §6: corporate_bs |
| Startup/VC | §7: startup_bs |
| Political | §8: political_bs |
| HR/People | §9: hr_bs |
| Marketing | §10: marketing_bs |
/demerzel bs-decode --self runs the decoder against Demerzel's own governance artifacts. This is critical — the grammar itself warns:
Self-check: Is Demerzel creating governance artifacts to govern governance artifacts that govern governance? If yes, this IS the BS.
All test results are FuzzyEnum values, not discrete. A statement might be:
specificity: { High: 0.3, Medium: 0.6, Low: 0.1 }
This means 60% medium specificity, 30% high, 10% low. The weighted combination produces a FuzzyEnum with membership across all five levels.
{
"input": "We are leveraging our next-generation AI platform to deliver unprecedented insights and drive transformative outcomes across the enterprise.",
"domain_match": "ai_bs (§2), corporate_bs (§6)",
"structural_analysis": {
"specificity": {
"result": { "Low": 0.85, "Medium": 0.15 },
"reasoning": "Replace 'our platform' with any company — sentence unchanged"
},
"falsifiability": {
"result": { "Zero": 0.7, "Low": 0.3 },
"reasoning": "No criteria for 'unprecedented' or 'transformative' — unfalsifiable"
},
"commitment": {
"result": { "Low": 0.9, "Medium": 0.1 },
"reasoning": "No who, no what specifically, no when"
}
},
"content_analysis": {
"density": {
"result": { "Fluff": 0.8, "Mixed": 0.2 },
"reasoning": "Remove 'next-generation', 'unprecedented', 'transformative', 'across the enterprise' — almost nothing remains"
},
"factual_density": {
"result": { "None": 0.95, "Sparse": 0.05 },
"reasoning": "Zero numbers, dates, names, or verifiable facts"
},
"actionability": {
"result": { "None": 0.9, "Vague": 0.1 },
"reasoning": "No one could act on this statement"
}
},
"rhetorical_analysis": {
"devices_found": [
{ "type": "loaded_language", "instances": ["next-generation", "unprecedented", "transformative"] },
{ "type": "strategic_ambiguity", "instances": ["across the enterprise"] },
{ "type": "weasel_words", "instances": ["drive...outcomes"] }
],
"evasion_tactics_found": [],
"rhetorical_density": 3.0
},
"bs_score": {
"weighted_raw": 0.12,
"classification": {
"Pure": 0.65,
"Serious": 0.30,
"Moderate": 0.05
},
"dominant": "Pure"
},
"tetravalent_mapping": "C",
"tetravalent_reasoning": "Statement sounds meaningful but conveys zero information — contradictory signal",
"rewrite": "We built [product name]. It [does X] for [Y users]. [Metric] improved by [N%]. Limitations: [list].",
"clear_speech_template": "claim + evidence + specifics + commitment"
}
grammars/gov-bs-generators.ebnf (Universal BS Generator & Detector Grammar v2), policies/scientific-objectivity-policy.yaml, constitutions/default.constitution.md Articles 1, 2, 5
Use when a pull request needs review by a different AI model than the one that authored it, or when configuring cross-model code review pipelines in CI/CD.
Run Demerzel's autonomous driver — full cycle or individual phases across all repos
Capture meaningful session state (current cursor, in-flight work, live hypotheses, open questions, do-NOT-carry-forward, success criteria) to state/digests/latest.md so the next session — including one after auto-compaction — can re-enter without re-discovering context cold. Distinct from /learnings (which captures surprises). Validates against docs/contracts/digest-schema.json.
Autonomous research scheduler — selects the most under-researched department by coverage ratio, generates questions from completeness gaps, cross-model validates, auto-produces courses at T>=0.8, rate-limited to 6 cycles/day with kill switch
Personalized AI tutor — turns any topic into a customized, progress-tracked learning course. Use when the user wants to LEARN a concept, skill, language, tool, or domain. Creates a MISSION, curated RESOURCES, iterative lessons with self-checks, and learning-records that persist across sessions. NOT for interrogating the user about their own design (use brainstorming/IDSD for that).
Run one bounded supervised autonomous cycle on the Demerzel governance repo. Picks the smallest unchecked governance slice inside allow_edit, implements it, runs the oracle (`pwsh scripts/verify.ps1`), emits cycle evidence, and stops. Refuses to run unless dev-process-overseer reports loop-eligible and the preflight passes. Use when asked to "run one supervised loop", "advance governance autonomously", or "do one cycle and stop".