🛰️ Year Ahead

Generates the deepest scheduled forward-look at Swedish politics — a 365-day annual outlook anchored in the freshest IMF WEO vintage available at run time (April or October), tracking the Swedish budget rhythm (BP autumn + VP spring), the Riksmöte calendar, and EU presidency rotations affecting Swedish politics. Tier-C aggregation × 2.0 depth multiplier.

The dedicated news-translate workflow runs on a separate track and translates executive-brief.md markdown into 13 language siblings (executive-brief_<lang>.md)

What this workflow does

• Article type: year-ahead (registry id; see analysis/article-types.json)
• Analysis subfolder: analysis/daily/$ARTICLE_DATE/year-ahead/
• Aggregated markdown: analysis/daily/$ARTICLE_DATE/year-ahead/article.md
• Rendered HTML: news/$ARTICLE_DATE-year-ahead-{en,sv,da,no,fi,de,fr,es,nl,ar,he,ja,ko,zh}.html — always all 14 languages
• Horizon: 365 days; lookback 180 days.
• Single-run model: download → analysis Pass 1 + 2 → gate → aggregate → render (14 languages) → ONE PR.

Long-horizon mandate (from ext/long-horizon-forecasting.md)

• Scenario count: ≥ 4 base scenarios + 5 wildcards in scenario-analysis.md + wildcards-blackswans.md (the supplementary wildcards-blackswans template is blocking for this article type).
• Counterfactuals: ≥ 2 explicit counterfactual paragraphs in devils-advocate.md.
• Cross-horizon citations: cite ≥ 2 most recent quarter-ahead analyses + ≥ 4 most recent monthly-review analyses in cross-reference-map.md. Missing citations fail the gate.
• PESTLE mandatory: pestle-analysis.md (otherwise supplementary) is blocking — outside-in environmental scan across all six dimensions.
• Quantitative-SWOT mandatory: quantitative-swot.md is blocking — every SWOT row carries a numeric weighting and impact estimate.
• IMF policy: full imf-fetch.ts compare across Nordic peers (SWE, DNK, NOR, FIN) for WEO macro + FM fiscal + DOTS bilateral trade. Each WEO/FM citation must include the projection-year stamp (T+1, T+2, T+5).
• Forward indicators: ≥ 12 dated indicators across month / quarter / year / cycle / election bands.
• Word floor: ≥ 2 500 words.
• Election cycle anchor: current (until 2026-09-13); flips to next after the election (registry-driven).

Time budget

🟡 Plan to call safeoutputs___create_pull_request by agent minute 42 (hard deadline 45) to reserve job-level headroom for setup variance and the safe-outputs runner. See 00-base-contract.md §Session timing and 07-commit-and-pr.md §Deadline enforcement.

🔴 Token budget awareness: This workflow uses claude-opus-4.8 which consumes tokens rapidly on complex analysis. The 25M token budget can be exhausted in ~20 minutes of intensive MCP querying + large file writes. Check agent_minute before EVERY phase transition. If agent_minute ≥ 20 and zero analysis artifacts exist on disk, immediately compress scope to a minimal viable set and target PR by minute 35.

Minutes	Phase
0–3	MCP pre-warm + IMF vintage pin
3–7	Download data (Riksdag + SCB + IMF Nordic-peer compare)
7–27	Analysis Pass 1 (all 23 artifacts + PESTLE + wildcards + quantitative-SWOT at 2.0× depth)
27–35	Analysis Pass 2 (read-back; counterfactuals; horizon-band stratification)
35–37	Analysis Gate (long-horizon checks)
37–38	Aggregate (article.md)
38–40	Render (scripts/render-articles.ts --lang all → all 14 HTML)
40–42	Stage + commit + ONE safeoutputs___create_pull_request — HARD DEADLINE agent minute 45

🟡 Scope-compression rule: if you reach agent minute 35 without Pass 2 complete, halt Pass 2 deepening and run the gate against whatever you have — if-no-changes: warn will not silently fail the run, but a missing PR will. Always trim depth before iterating.

⚠️ HARD FILE LIMIT (200 files): The safe-outputs handler hard-rejects PRs with > 200 files (E003). You MUST run the 200-file guard from 07-commit-and-pr.md before calling safeoutputs___create_pull_request. Budget: 23 core artifacts + README (1) + article.md (1) + 14 HTML + pir-status.json (1) ≈ 40 files. Never stage documents/ or pass1/ directories. If staged count exceeds 180, unstage documents/ then JSON files until under budget. This is non-negotiable — the previous run failed with 269 files.

Inputs

• article_date — override date (defaults to today)
• force_generation — regenerate even if recent year-ahead exists (within 60 days)
• analysis_depth — defaults to comprehensive

Note: there is no languages input. Every run produces all 14 language HTML files. Translation depth-of-quality scales with the time budget (see the table above).

File-write contract

🛠 Write every analysis artifact (analysis/daily/$DATE/$SUB/*.md, documents/*.md, JSON sidecars, methodology-reflection.md re-run deltas) with the edit tool. cat <<'QUOTED_EOF' > file is a Tier-2 fallback only — ASCII-only, no code fences / Mermaid / $ / backticks / EOF markers, < 200 lines, and only after edit has failed for a non-content reason. Banned for analysis/daily/** writes: python3, node -e, sed -i, echo … > file, tee file, unquoted heredocs (<<EOF) — sole exception: the pre-flight scaffold in 03-data-download.md (env-var refs and short literals only, no agent-generated content). The aggregator (scripts/aggregate-analysis.ts) and renderer (scripts/render-articles.ts) are the only writers for article.md and news/*.html. See 01-bash-and-shell-safety.md §File creation & overwrite strategy.

All other rules live in the imported modules.