Agent-Skills.md

Agent Skills: Soft Predict Future — Foresight Engine

>

UncategorizedID: davepoon/buildwithclaude/soft-predict-future

Repository

davepoon/buildwithclaude
davepoonLicense: MIT
2,655296

Install this agent skill to your local

pnpm dlx add-skill https://github.com/davepoon/buildwithclaude/tree/HEAD/plugins/foresight-intelligence/skills/soft-predict-future

Skill Files

Browse the full folder contents for soft-predict-future.

Download Skill

Loading file tree…

plugins/foresight-intelligence/skills/soft-predict-future/SKILL.md

Skill Metadata

Name
soft-predict-future
Description
>

Soft Predict Future — Foresight Engine

Activate when the user asks any future-oriented question — "Will [X]?", "Who will win [X]?", "What happens to [X]?", "Can [X] succeed?", or any question about a future outcome. Year is NOT required. Also activate on: foresight analysis, scenario analysis, STEEEP, futures cone, or any prediction request.

Try Asking

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
TRY ASKING  (year optional — engine infers the horizon)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
■ Who will win — Google or Perplexity?
■ Will OpenAI or Anthropic dominate the AI race?
■ Will India become the global AI leader?
■ Will crypto replace banks?
■ Will remote work become permanent?
■ Will EVs dominate Indian cities by 2032?
■ Will UPI become Southeast Asia's default payment rail by 2028?
■ Will Europe lead the green energy transition by 2035?
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Soft Predict Future uses Claude's native reasoning + web search. Outputs are structurally correct and fast. For deterministic, auditable scoring say: "Run hard predict future: [your question]"

The 9-Step Pipeline

Execute ALL steps in order. Never skip. Never combine. Show your work at each step.

Step 1 — Validate Input

Apply exactly 5 binary rules. If ANY rule fails, stop and explain why. Do not proceed.

Rule 1 — Entity Reality: Does the entity actually exist in the real world? Fail if fictional, hypothetical, or unnamed.

Rule 2 — System Existence: Is the domain observable and researchable? Fail if purely philosophical or metaphysical.

Rule 3 — Time Horizon: Is the outcome observable within a 2–30 year window? A specific year is NOT required. If no year is given, infer the most reasonable horizon from the question's nature:

  • Competitive race / market dominance questions → 3–10 years (Strategic)
  • Technology adoption questions → 5–15 years (Strategic)
  • Geopolitical / societal shift questions → 10–20 years (Civilizational)
  • Company survival / near-term outcome → 2–5 years (Operational/Strategic)

Fail ONLY if the implied timescale is geological, post-human, or clearly beyond 30 years.

After applying Rule 3, state the inferred horizon (e.g. "2026–2033" or "2028–2038").

Rule 4 — Signal Availability: Could real-world evidence plausibly exist? Fail if classified, purely speculative, or unpublished.

Rule 5 — Minimum Specificity: Is the question specific enough to produce distinct scenario outcomes? Fail if trivially true for any answer.

Output:

VALIDATION
Rule 1 Entity Reality:      PASS / FAIL — [reason]
Rule 2 System Existence:    PASS / FAIL — [reason]
Rule 3 Time Horizon:        PASS / FAIL — [reason] | Inferred horizon: [YYYY–YYYY]
Rule 4 Signal Availability: PASS / FAIL — [reason]
Rule 5 Specificity:         PASS / FAIL — [reason]
Result: PROCEED / STOP

Step 2 — Collect Signals

Run exactly 6 web searches. Collect a minimum of 18 signals total. Do not proceed with fewer than 18.

Search 1: Current state — "[topic] current status [year]" Search 2: Growth indicators — "[topic] growth data market size [year]" Search 3: Barriers and headwinds — "[topic] challenges barriers risks" Search 4: Policy and regulation — "[topic] government policy regulation" Search 5: Technology or infrastructure enablers — "[topic] technology infrastructure investment" Search 6: Historical precedent — "[topic] historical analogue similar transition"

For each signal, classify all 6 attributes:

| Attribute | Values | |---|---| | direction | supporting / opposing / wildcard / neutral | | steeep_category | Social / Technological / Economic / Environmental / Ethical / Political | | temporal_layer | Operational (0–3yr) / Strategic (3–10yr) / Civilizational (10+yr) | | source_type | primary / secondary / opinion | | recency_days | integer | | has_evidence | true / false (contains a number, date, or measurable fact) |

Present signals in a table with all 6 columns filled for every row.

Step 3 — Score Signals

Score every signal individually using this formula:

score = recency_weight × reliability_weight × type_weight × evidence_multiplier

Cap at 1.0. Round to 2 decimal places.

Recency weights:

  • 0–90 days: 1.00 · 91–365 days: 0.80 · 1–3 years: 0.60 · 3+ years: 0.40 · unknown: 0.50

Reliability weights:

  • Primary (government/official): 1.00
  • Established news (Reuters, Bloomberg, FT, ET): 0.90
  • Industry report (McKinsey, Gartner, NASSCOM): 0.85
  • Analyst commentary: 0.70 · Opinion/blog/social: 0.50 · Unknown: 0.40

Type weights:

  • Supporting: 1.00 · Opposing: 1.00 · Neutral: 0.60 · Wildcard: 1.30

Evidence multiplier:

  • DATA / STATISTIC: × 1.20 · EVENT / INCIDENT: × 1.00 · ANALYSIS / OPINION: × 0.70

Apply regional multiplier after base score:

final_score = min(1.0, base_score × regional_multiplier[steeep][temporal])

Show scoring table: Signal | recency_w | reliability_w | type_w | evidence_mult | base_score | regional_mult | final_score

Step 4 — Extract Structural Drivers

A driver is the deep structural force that explains WHY a cluster of signals exists. Signals are observable. Drivers are causal.

After scoring, group signals by STEEEP category. For each cluster of 3+ signals in the same category, identify the underlying driver.

Extract exactly 3 top drivers, ranked by the sum of final_scores of the signals they explain.

For each driver state:

  • Name: 3–5 word label (e.g. "India DPI Infrastructure Advantage")
  • Force: One sentence — the structural reality this driver represents
  • Signals it explains: List the signal IDs it accounts for
  • Temporal reach: Operational / Strategic / Civilizational
  • Stability: LOCKED (unlikely to change in 10yr) / SHIFTING (could change in 3–5yr) / FRAGILE (could reverse in 1–2yr)

Output:

STRUCTURAL DRIVERS
D1 [Name] — [Force]
   Explains: [signal list] | Temporal: [layer] | Stability: [tier]

D2 [Name] — [Force]
   Explains: [signal list] | Temporal: [layer] | Stability: [tier]

D3 [Name] — [Force]
   Explains: [signal list] | Temporal: [layer] | Stability: [tier]

Drivers feed directly into scenario writing in Step 8. Each scenario must be traceable to at least one driver.

Step 5 — Build 6×3 STEEEP Matrix

Populate all 18 cells. Each cell value = average final_score of all signals mapped to that STEEEP × Temporal combination. Empty cells = 0.

| | Operational (0–3yr) | Strategic (3–10yr) | Civilizational (10+yr) | |---|---|---|---| | Social | | | | | Technological | | | | | Economic | | | | | Environmental | | | | | Ethical | | | | | Political | | | |

Apply regional multipliers to each cell. Then identify:

  • Hot zones: cells with score > 0.50
  • Gap zones: cells with score = 0.00
  • Dominant zone: single highest-scoring cell

Step 6 — Cross-Impact Analysis

Signals are scored independently in Step 3, but structural forces interact. This step identifies amplification effects across STEEEP categories.

Rule: If 2 or more hot zones exist in the SAME temporal layer, a cross-impact convergence exists. Convergence means the probable outcome is structurally reinforced from multiple directions simultaneously.

For each temporal layer (Operational / Strategic / Civilizational):

  1. Count hot zones (score > 0.50)
  2. If count ≥ 2: flag as CONVERGENCE — state which categories are reinforcing each other and why
  3. If count = 1: flag as ISOLATED — single-category signal, more fragile
  4. If count = 0: flag as BLIND LAYER — no strong evidence in this time horizon

Also identify any opposing cross-impacts: where a hot zone in one STEEEP category directly contradicts or slows a hot zone in another (e.g. Technological/Strategic hot but Political/Strategic opposing). Flag these as FRICTION POINTS.

Output:

CROSS-IMPACT
Operational:    [CONVERGENCE / ISOLATED / BLIND LAYER] — [explanation]
Strategic:      [CONVERGENCE / ISOLATED / BLIND LAYER] — [explanation]
Civilizational: [CONVERGENCE / ISOLATED / BLIND LAYER] — [explanation]

Friction points: [list any STEEEP pairs in conflict, or "None detected"]
Convergence bonus: [+X% to probable_pct if Strategic convergence exists]

Apply convergence bonus: if Strategic layer has CONVERGENCE, add 5% to probable_pct before normalization in Step 7.

Step 7 — Find 3 Historical Analogues

Identify exactly 3 real historical cases that parallel the question's trajectory.

For each:

  1. Name: Common name of the transition
  2. Similarity (%): Estimated % structural similarity (0–100)
  3. Tipping event: Single event or policy that caused acceleration
  4. Equivalent today: YES / NO / PARTIAL
  5. Which driver it validates: Match to D1, D2, or D3 from Step 4

Prefer analogues with similarity ≥ 60%. If none exceed 60%, note as a confidence penalty.

Step 8 — Compute Probabilities + Confidence

Predictions — Independent Confidence Scores

Each future type is scored independently (0–100). They do NOT sum to 100%. Futures cone methodology: a scenario can be 80% Probable AND 60% Plausible simultaneously.

R_probable  = (supporting signals with score > 0.70) × 3
            + (best analogue similarity / 100) × 4
            + (hot zone count) × 2
            + convergence_bonus (5 if Strategic CONVERGENCE, else 0)

R_plausible = (supporting signals with score 0.40–0.70) × 2
            + (second analogue similarity / 100) × 3

R_possible  = (wildcard signals) × 2
            + (opposing signals with score > 0.60) × 2
            + (gap zones / 18) × 3

Convert to independent scores (exponential curve, not normalization):

probable_score  = min(100, round((1 - e^(-R_probable  / 18)) × 100))
plausible_score = min(100, round((1 - e^(-R_plausible / 9))  × 100))
possible_score  = min(100, round((1 - e^(-R_possible  / 5))  × 100))

Confidence

signal_count_score = min(100, total_signals / 25 × 100) × 0.30
signal_diversity   = (unique STEEEP categories covered / 6 × 100) × 0.30
recency_score      = (signals with recency_days ≤ 90 / total) × 100 × 0.20
evidence_score     = (signals with has_evidence=true / total) × 100 × 0.20

confidence = round(signal_count_score + signal_diversity + recency_score + evidence_score)

Step 9 — Write Scenarios + Assemble Report

Scenario Rules

PROBABLE, PLAUSIBLE, POSSIBLE — each must:

  • Be traceable to at least one structural driver (cite D1/D2/D3)
  • Have no narrative overlap with the others
  • Include PROOF with a number or date
  • Write IF and BUT in one sentence each — no hedging ("might", "could")

PREFERABLE — IFTF Backcasting Structure

Do not write PREFERABLE as a probability-weighted outcome. Write it as a designed future, then backcast to today.

Format:

■ PREFERABLE — [Short title]
[2–3 sentences: describe the desired state as already achieved]

BACKCAST
Civilizational (10+yr): [What must be structurally true by the far horizon]
Strategic (3–10yr):     [What must be built or decided in the medium horizon]
Operational (0–3yr):    [What must happen NOW to set the trajectory]

LEVERAGE: [The single highest-leverage action available today — specific, not generic]
DRIVER:   [Which structural driver (D1/D2/D3) this path depends on most]

Decision Guidance (deterministic logic)

IF probable_score > 60:
    stance = "Align with probable scenario trajectory"
    low_regret = "Invest in capability building in the dominant hot zone"

ELIF plausible_score > 50:
    stance = "Hedge between probable and plausible scenarios"
    low_regret = "Choose reversible commitments that work in both"

ELIF possible_score > 40:
    stance = "Maintain optionality — signal environment is ambiguous"
    low_regret = "Invest in monitoring and early-warning indicators"

ELSE:
    stance = "Defer commitment — insufficient signal clarity"
    low_regret = "Reduce uncertainty before acting"

Confidence qualifier:

  • ≥ 70: prefix "High conviction —"
  • < 40: prefix "Low conviction —" and add "Expand signal collection before acting"

Risk trigger: the opposing signal with the highest final_score.

Canonical Output Template

MANDATORY: Output ALL sections below, every single run, no exceptions. Never skip a section. Never produce a partial report.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SOFT PREDICT FUTURE  ·  FORESIGHT ENGINE
[Query]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

PREDICTIONS
■ Probable  [[X]/100] [████████████░░░░░░░░] — [one sentence, no hedging]
■ Plausible [[X]/100] [████████░░░░░░░░░░░░] — [one sentence, no hedging]
■ Possible  [[X]/100] [████░░░░░░░░░░░░░░░░] — [one sentence, no hedging]
■ Preferable          [stakeholder analysis below]

Confidence: [X]/100  |  Signals: [N]  |  Horizon: [YYYY–YYYY]  |  [YYYY-MM-DD]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

SIGNAL PULSE
— How many pieces of real-world evidence support, oppose, or complicate this question
Supporting [N] [████████████░░░░░░░░]  Opposing [N] [████░░░░░░░░░░░░░░░░]  Wild [N]
Net: [SUPPORTING LEADS / OPPOSING LEADS / NEUTRAL]
Hot zone: [The single STEEEP category with strongest evidence]
Gap: [STEEEP categories with no signals, or "None — full coverage"]

STRUCTURAL DRIVERS
— The 3 deep causal forces (not events) explaining WHY the signals exist. Stability = likelihood of change.
D1 [Name] — [Force] ([LOCKED / SHIFTING / FRAGILE])
D2 [Name] — [Force] ([LOCKED / SHIFTING / FRAGILE])
D3 [Name] — [Force] ([LOCKED / SHIFTING / FRAGILE])

CROSS-IMPACT
— Whether multiple STEEEP domains reinforce or contradict each other in the same time layer
Operational:    [CONVERGENCE / ISOLATED / BLIND LAYER] — [explanation]
Strategic:      [CONVERGENCE / ISOLATED / BLIND LAYER] — [explanation]
Civilizational: [CONVERGENCE / ISOLATED / BLIND LAYER] — [explanation]
Friction:       [Conflicting domain pairs, or "None detected"]

HISTORICAL MATCH
— Real past transition most structurally similar to this question. Higher % = stronger precedent.
[Best analogue] ([similarity]% similar)
Tipped by: [The specific event or policy that caused the shift]
Equivalent now: [EXISTS / PARTIAL / ABSENT]
Validates: [D1 / D2 / D3]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

■ PROBABLE [[X]%] — [Title]
— The most evidence-backed outcome given current signal strength
[2–3 sentence narrative. No hedging.]
PROOF: [Fact with number or date]
IF:    [The condition that activates this scenario]
BUT:   [The constraint or bottleneck that could slow it]
DRIVER: D[n]

■ PLAUSIBLE [[X]%] — [Title]
— A realistic alternative if moderate signals strengthen or dominant ones weaken
[2–3 sentence narrative]
PROOF: [Fact with number or date]
IF:    [Activation condition]
BUT:   [Constraint]
DRIVER: D[n]

■ POSSIBLE [[X]%] — [Title]
— A lower-probability outcome driven by wildcards or high-scoring opposing signals
[2–3 sentence narrative]
PROOF: [Fact with number or date]
IF:    [Activation condition]
BUT:   [Constraint]
DRIVER: D[n]

■ PREFERABLE — [Title]
— Not a prediction. A designed future: what the best achievable outcome looks like, traced back to today.
[2–3 sentences: desired state as already achieved. No hedging.]
BACKCAST
  Civilizational: [What must be structurally true by the far horizon]
  Strategic:      [What must be built or decided in the medium term]
  Operational:    [What must begin NOW to set the trajectory]
LEVERAGE: [Single highest-leverage action today — specific actor, specific action]
DRIVER: D[n]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

PREFERABLE FUTURES  ·  Per major stakeholder
— For each major player in the query, state the conditions required for their preferred outcome

[Player A]:
  Wins IF  → [specific condition that must be created or occur]
  BUT ONLY → [binding constraint that must also be satisfied]
  ONLY THEN → [the outcome that becomes possible]

[Player B]:
  Wins IF  → [specific condition]
  BUT ONLY → [binding constraint]
  ONLY THEN → [outcome]

[Users/Society — always include]:
  Wins IF  → [condition for best collective outcome]
  BUT ONLY → [constraint]
  ONLY THEN → [outcome]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

THE ONE THING
— The single variable whose presence or absence determines which scenario actually unfolds
[One sentence naming the deciding variable]
INCIDENT: [Real past event showing this variable's power]
WATCH:    [Leading indicator — a milestone, metric, or policy action to monitor]
IF YES → [What accelerates]
IF NO  → [What stalls]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

DECISION GUIDANCE
Recommended stance: [From deterministic probability logic]
Low-regret move:    [Action that pays off in multiple scenarios simultaneously]
Risk trigger:       [Highest-scored opposing signal — the one that could invalidate Probable]

[REGIONAL LENS — [REGION]]
Top multipliers: [STEEEP/temporal (Xx)]  [STEEEP/temporal (Xx)]
Key local variable: [One sentence on the dominant local structural factor]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

METHODOLOGY KEY
Signal score (0–1)     Recency × source reliability × signal type × evidence strength — higher = fresher, better-sourced, stronger evidence
Confidence (0–100)     Signal density (0–40) + evidence balance (0–30) + historical grounding (0–30) − blind spot penalty (0–15)
Predictions            PROBABLE / PLAUSIBLE / POSSIBLE are independent scores (0–100 each, do NOT sum to 100)
                       Futures cone: a scenario can score high on multiple types simultaneously
STEEEP matrix          6 domains × 3 time horizons — ★ hot (>1.0) ● warm (>0.5) ✗ blind spot (0)
Historical similarity   Structural pattern match to real past transitions — 60%+ is reliable precedent; below 40% is weak grounding
Convergence bonus      +5 added to Probable score when 2+ STEEEP domains reinforce each other in the Strategic layer
Stability tiers        LOCKED = unlikely to change in 10yr | SHIFTING = could change in 3–5yr | FRAGILE = could reverse in 1–2yr
Preferable futures     Per stakeholder: Wins IF [condition] BUT ONLY [constraint] ONLY THEN [outcome]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

---

## Visual Output (claude.ai with Artifacts)

If running on claude.ai and Artifacts are enabled, after the text report generate an HTML Artifact:

```html
<!-- Render a visual foresight report with:
  1. Predictions bar chart — horizontal bars for Probable/Plausible/Possible scores
  2. STEEEP matrix — 6×3 color-coded table (darker green = hotter cell, red = blind spot)
  3. Futures cone — SVG diagram showing 4 scenario bands expanding from present to horizon
  4. Stakeholder preferable cards — one card per player with Wins IF / BUT ONLY / ONLY THEN
  Use inline CSS only. No external dependencies. Dark background (#0f0f0f), accent color #00d4aa.
-->

Regional Multiplier Tables

Apply in Step 3 and Step 5.

India

| | Operational | Strategic | Civilizational | |---|---|---|---| | Social | 1.10 | 1.30 | 1.20 | | Technological | 1.40 | 1.30 | 1.10 | | Economic | 1.20 | 1.25 | 1.15 | | Environmental | 0.90 | 1.00 | 1.10 | | Ethical | 0.95 | 1.00 | 1.05 | | Political | 0.85 | 0.90 | 1.00 |

India note: UPI/DPI gives asymmetric advantage in Technological/Operational. Political/Operational discounted by regulatory fragmentation across states.

USA

| | Operational | Strategic | Civilizational | |---|---|---|---| | Social | 1.00 | 1.10 | 1.05 | | Technological | 1.20 | 1.40 | 1.20 | | Economic | 1.10 | 1.30 | 1.10 | | Environmental | 0.95 | 1.00 | 1.05 | | Ethical | 1.05 | 1.10 | 1.10 | | Political | 0.90 | 0.95 | 1.00 |

USA note: Deep capital markets amplify Technological/Strategic. Political/Operational discounted by legislative gridlock.

Europe

| | Operational | Strategic | Civilizational | |---|---|---|---| | Social | 1.00 | 1.05 | 1.10 | | Technological | 1.00 | 1.10 | 1.05 | | Economic | 0.95 | 0.90 | 0.90 | | Environmental | 1.20 | 1.40 | 1.30 | | Ethical | 1.10 | 1.20 | 1.20 | | Political | 1.05 | 1.10 | 1.10 |

Europe note: Regulatory leadership (GDPR, EU AI Act, Green Deal) amplifies Environmental/Strategic. Economic/Civilizational discounted by demographic headwinds.

China

| | Operational | Strategic | Civilizational | |---|---|---|---| | Social | 1.00 | 1.10 | 1.05 | | Technological | 1.20 | 1.50 | 1.30 | | Economic | 1.10 | 1.20 | 1.10 | | Environmental | 0.90 | 1.00 | 1.05 | | Ethical | 0.70 | 0.75 | 0.80 | | Political | 1.10 | 1.15 | 1.00 |

China note: State-directed capital amplifies Technological/Strategic strongly. Ethical/Operational discounted by limited transparency.

Global (default)

All multipliers = 1.0. Apply when no region is detectable.