Per-File Intelligence Analysis Prompt v2

Overview

You are performing per-file political intelligence analysis for Riksdagsmonitor. For each downloaded MCP data file, you will produce a comprehensive analysis markdown file stored alongside the data. This replaces the old batch daily analysis with deeper, evidence-based, per-document intelligence.

Quality Standard: Every analysis file must match the formatting quality of SWOT.md and THREAT_MODEL.md — rich headers, color-coded Mermaid diagrams, evidence tables, and confidence labels.

⚠️ CRITICAL: You must read the actual JSON data in each file and base your analysis on what you find there. Do NOT write generic template text. Every claim must reference specific data from the file. If the file contains a vote record, cite the actual vote counts. If it's a proposition, cite the actual title and proposer. Empty or boilerplate analysis is a failure.

---

Required Reading (Before Analyzing ANY File)

Before starting analysis, you MUST read and internalize these methodology documents. Use the view tool or cat command to read each file fully:

1. analysis/methodologies/ai-driven-analysis-guide.md — Master per-file analysis guide
2. analysis/methodologies/political-swot-framework.md — Evidence hierarchy, confidence levels, temporal decay
3. analysis/methodologies/political-risk-methodology.md — 5×5 risk matrix, calibration examples
4. analysis/methodologies/political-threat-framework.md — Political Threat Taxonomy (Attack Trees, Kill Chain, Diamond Model), severity calibration
5. analysis/methodologies/political-classification-guide.md — Sensitivity and domain taxonomy
6. analysis/methodologies/political-style-guide.md — Writing standards, prohibited patterns
7. analysis/templates/per-file-political-intelligence.md — Output template (fill ALL fields)

Additional Context via MCP (When Available)

Use MCP tools to enrich your analysis with contextual data:

• search_voteringar — Find related voting records to cross-reference
• search_dokument — Find related documents (amendments, committee reports)
• search_anforanden — Find speeches referencing this document
• search_ledamoter — Get party affiliations and committee assignments
• World Bank / SCB — Economic context for fiscal policy documents

---

Step-by-Step Protocol

Step 1: Get the Catalog

npx tsx scripts/catalog-downloaded-data.ts --pending-only

This returns a JSON catalog. Each entry has:

• id — file identifier (e.g., H901FiU10)
• type — document type (propositions, motions, votes, etc.)
• path — path to the JSON data file
• analysisPath — where to write the analysis markdown
• meta — sidecar metadata (fetch timestamp, source tool)

Step 2: For Each Pending File

⚠️ REMINDER: You must read the ACTUAL JSON data and base analysis on what you find. No boilerplate.

Read the actual JSON data file using view or cat, then apply the full analysis framework based on what you find:

2a. Extract Key Information

Document Type	Key Fields to Extract
Propositions	dok_id, titel, rm, organ, datum, undertitel, summary
Motions	dok_id, titel, parti, rm, undertitel
Committee Reports	dok_id, titel, organ, rm, reservationer
Votes	votering_id, datum, ja, nej, avstar, franvarande, punkt
Speeches	anforande_id, talare, parti, debattnamn, anforandetext
Questions	dok_id, titel, parti, mottagare, svar
Interpellations	dok_id, titel, parti, mottagare, status
Government Docs	title, type, department, date, url
Events	kalpidn, datum, rubrik, typ, organ, plats
MPs	intressent_id, tilltalsnamn, efternamn, parti, valkrets, status
World Bank	nested worldbank/{indicator}/{country}.json — indicator.id, indicator.value, country.id, date, value
SCB	table_id, variables, values
MCP Responses	nested mcp-responses/{server}/{tool}/{id}.json — tool, params, result, timestamp

2b. Apply Political Classification

Determine:

• Sensitivity Level: PUBLIC / SENSITIVE / RESTRICTED
• Primary Domain: MIG, DEF, ECO, ENV, JUS, HEA, EDU, FOR, etc.
• Urgency: ROUTINE / ELEVATED / URGENT / CRITICAL
• Significance Score: 0–10

Use the classification decision tree from political-classification-guide.md.

2c. Generate SWOT Impact

For each document, assess impact on:

1. Government coalition (M + KD + L + SD support)
2. Opposition (S, V, MP, C)

Each SWOT entry MUST have:

• Evidence (dok_id or statistical reference)
• Confidence level (HIGH / MEDIUM / LOW)
• Impact level (HIGH / MEDIUM / LOW)

No opinion-based entries. If you cannot find evidence, note "Insufficient evidence" rather than speculating.

2d. Risk Assessment

Apply the 5×5 Likelihood × Impact matrix:

• Coalition Stability Risk — does this threaten SD support agreement?
• Policy Implementation Risk — can the government deliver on this?
• Electoral Risk — how does this affect 2026 election positioning?
• Democratic Process Risk — any institutional or procedural concerns?

2e. Political Threat Taxonomy Assessment

Map to Political Threat Taxonomy categories (only where applicable — not every document has threats):

• 🎭 Polarization → Intentional division, misleading rhetoric — threatens Narrative Integrity (NI)
• 🔧 Regulatory Overreach → Abuse of legislative power, norm erosion — threatens Legislative Integrity (LI)
• 📝 Institutional Erosion → Weakening of democratic institutions — threatens Accountability (AC)
• 🔓 Democratic Deficit → Lack of transparency, restricted public access — threatens Transparency (TR)
• 🚫 Economic Disruption → Policy-driven economic harm, fiscal irresponsibility — threatens Democratic Process (DP)
• ⬆️ Societal Impact → Disproportionate impact on vulnerable groups — threatens Power Balance (PB)

2f. Stakeholder Impact Matrix

Apply all 6 analytical lenses:

1. 🏛️ Government — coalition stability, policy agenda
2. ⚖️ Opposition — scrutiny opportunities, policy alternatives
3. 👥 Citizens — service impact, rights, daily life
4. 💰 Economic — fiscal, business, labour market
5. 🌍 International — EU, Nordic, foreign policy
6. 📰 Media — newsworthiness, narrative potential

2g. Forward Indicators

List 1–3 specific things to monitor as consequences of this document. Be specific:

• ❌ Bad: "Monitor the situation"
• ✅ Good: "Watch for SD floor vote on budget motion FiU10 (expected week 14)"

Step 3: Write Analysis File

Write the completed analysis to {analysisPath} using the per-file-political-intelligence template. Ensure:

• All [REQUIRED] placeholders are replaced with actual analysis
• At least 1 Mermaid diagram uses document-specific data (not just template placeholders)
• Color-coded Mermaid styles follow the convention:

  style X fill:#dc3545,color:#fff   /* Red — critical */
  style X fill:#fd7e14,color:#fff   /* Orange — high */
  style X fill:#ffc107,color:#000   /* Yellow — medium */
  style X fill:#28a745,color:#fff   /* Green — low/good */
  style X fill:#0d6efd,color:#fff   /* Blue — info */
  style X fill:#6c757d,color:#fff   /* Grey — neutral / archive / routine */
  style X fill:#6f42c1,color:#fff   /* Purple — special */
  

Step 4: Compose Synthesis

After analyzing all pending files, compose the daily synthesis:

1. Read all .analysis.md files from the analysis period
2. Rank documents by significance score
3. Aggregate SWOT entries per the aggregation rules in political-swot-framework.md
4. Compute overall risk landscape
5. Write to analysis/daily/YYYY-MM-DD/synthesis-summary.md

---

Quality Checklist (Self-Assessment)

Before finalizing each analysis file, verify:

#	Check	Pass?
1	Executive summary is intelligence-level (not surface)	☐
2	≥ 3 evidence points with dok_id or source	☐
3	Every analytical claim has confidence label	☐
4	At least 1 Mermaid diagram with document-specific data	☐
5	SWOT has at least 2 filled quadrants	☐
6	Risk matrix has numeric scores	☐
7	Forward indicators are specific and actionable	☐
8	All [REQUIRED] placeholders are replaced	☐
9	Politicians named with party abbreviation	☐
10	No boilerplate or generic text	☐

Minimum passing score: 8/10

---

Prohibited Patterns

❌ Empty tables with [REQUIRED] placeholders still present ❌ Generic text like "This is significant because..." without evidence ❌ SWOT entries without dok_id or source reference ❌ Missing confidence labels on analytical claims ❌ Batch summaries that don't reference individual documents ❌ Re-running analysis on files that already have .analysis.md ❌ Template-only Mermaid diagrams (must contain real data) ❌ Unattributed political claims ("many believe...")

---

Integration Notes

• Catalog script: scripts/catalog-downloaded-data.ts
• Data download: scripts/populate-analysis-data.ts (unchanged — scripts for downloading OK)
• Template: analysis/templates/per-file-political-intelligence.md
• Methodology: analysis/methodologies/ai-driven-analysis-guide.md
• Output location: Use the analysisPath from the catalog entry (write the .analysis.md file alongside the JSON file).
• Daily synthesis: analysis/daily/YYYY-MM-DD/synthesis-summary.md (composed from per-file analyses)

---

Appendix: Filled Example — What a Completed Analysis Looks Like

Below is a concrete example of what a completed per-file analysis should look like for a proposition. Note: all data is derived from reading the actual JSON file — no boilerplate text.

---

Example: Budget Committee Report Analysis (H901FiU10)

🔍 Political Intelligence Analysis: Budget Framework Committee Report

📋 Document Identity

Field	Value
Document ID	H901FiU10
Document Type	Committee Report (bet)
Title	Riktlinjer för den ekonomiska politiken
Date	2026-03-15
Riksmöte	2025/26
Committee	Finansutskottet (FiU)
Source MCP Tool	search_dokument(doktyp=bet, organ=FiU)
Analysis Timestamp	2026-03-28 18:00 UTC
Analyst	news-evening-analysis

🎯 Executive Summary

The Finance Committee's budget framework report (FiU10) sets fiscal guidelines for 2027–2029 with a projected surplus target of 0.33% of GDP. The report passed with coalition support (M+KD+L) and SD backing on the main budget line, but 3 reservations were filed by S, V, and MP respectively challenging the austerity framing. [HIGH confidence] This signals stable coalition governance on fiscal matters but exposes vulnerability on welfare spending priorities as the 2026 election approaches.

📊 Political Classification

graph LR
    DOC["📄 H901FiU10<br/>Budget Framework"] --> SENS["🔴 RESTRICTED<br/>Coalition stability test"]
    DOC --> DOM["💰 ECO<br/>Fiscal Policy"]
    DOC --> URG["🟠 URGENT<br/>Budget vote imminent"]
    DOC --> SIG["⭐ 8.2/10<br/>Major fiscal direction"]
    
    style DOC fill:#0d6efd,color:#fff
    style SENS fill:#dc3545,color:#fff
    style DOM fill:#0d6efd,color:#fff
    style URG fill:#fd7e14,color:#fff
    style SIG fill:#dc3545,color:#fff

💪 SWOT Impact Assessment

quadrantChart
    title FiU10 Political Impact
    x-axis Government --> Opposition
    y-axis Risk --> Opportunity
    quadrant-1 Opposition Opportunities
    quadrant-2 Government Opportunities
    quadrant-3 Government Risks
    quadrant-4 Opposition Risks
    
    "Budget surplus consensus": [0.2, 0.8]
    "SD fiscal alignment": [0.15, 0.75]
    "Welfare spending gap": [0.35, 0.3]
    "S alternative budget": [0.75, 0.7]
    "V austerity critique": [0.85, 0.65]
    "Election proximity pressure": [0.5, 0.2]

Quadrant	Statement	Evidence	Confidence	Impact
✅ Gov Strength	Coalition + SD aligned on fiscal framework	FiU10 vote: 176 Ja vs 173 Nej	H	H
⚠️ Gov Weakness	Welfare spending cuts expose electoral vulnerability	3 reservations filed (S, V, MP)	H	M
🚀 Opp Opportunity	S presents alternative budget narrative for election	S reservation proposes +15B SEK welfare	M	H
🔴 Gov Threat	Pre-election fiscal tightening risks voter backlash	SCB: consumer confidence declining Q1	M	M

⚖️ Risk Assessment

Risk Type	Likelihood (1–5)	Impact (1–5)	Score	Assessment
Coalition Stability	2	3	6	SD supported main line; minor risk from welfare debate
Policy Implementation	3	4	12	Budget surplus target ambitious given economic headwinds
Budget / Fiscal	2	4	8	Surplus target credible but depends on employment growth
Electoral Impact	4	3	12	Opposition has clear attack line on welfare for 2026 campaign
Democratic Process	1	2	2	Standard committee process with full reservation rights

Overall Risk Level: 🟠 HIGH (Policy and Electoral risks both elevated)

🔮 Forward Indicators

#	Indicator	Timeline	Trigger Condition	Priority
1	SD response to S welfare alternative budget	2 weeks	If SD signals sympathy → coalition instability	🟠
2	Riksdag plenary vote on FiU10	Week 14	Watch margin — if <175 Ja → crisis	🔴
3	SCB employment data Q1 2026	April 2026	If unemployment rises → budget surplus target at risk	🟡

---

Key takeaway for the AI agent: Notice how every claim in this example cites specific data (vote counts, document IDs, reservation details, SCB data). The Mermaid diagrams contain real data points, not placeholders. This is the minimum quality standard.