v1.3.0 — universal scale + qualitative input + B/T currency

v1.3.0 — universal scale (no caps) + qualitative input per dimension + B/T currency

Cambios principales

1. Universal scale — sin cap de industria, tamaño ni revenue

• El instructions del MCP server ya no dice "serving $1M-$10M businesses". Ahora dice "ANY size worldwide, from solo founders to Fortune 500, revenue from $0 to $1B+, every industry from construction to aerospace".
• Los midpoints del _estimate_monthly_leak se extendieron: ahora acepta 10m-100m, 50m-100m, 100m-500m, 100m-1b, 500m-1b, 1b+, over100m, over1b, etc.
• El ICP outbound (ProEléctricos, healthcare, etc.) sigue siendo filtro COMERCIAL de David, pero ya NO es límite del SISTEMA.

2. Qualitative input por dimensión (11 campos nuevos)

• Cada una de las 11 dimensiones acepta un campo qual_<dim> opcional donde el cliente describe sus EMOCIONES y THOUGHTS sobre esa área (200-500 chars recomendado).
• Los qual inputs se persisten en dimensions[].qualitative_input del report.
• Enriquece el leadership_insight con datos cualitativos que las métricas puras no capturan (mindset, moral, bloqueos ocultos, miedos no dichos).
• Backward-compatible: si el campo viene vacío, la diagnosis corre igual que antes.

3. Currency formatting B/T (billones/trillones)

• El fmt() del roi_projector ahora soporta:
  • $1.2K (miles)
  • $5.4M (millones)
  • $2.3B (billones)
  • $1.5T (trillones)
• Localización adicional por agregar (mil/mill/billón/trillón) en v1.3.1.

Archivos modificados

• src/cia_diagnose/server.py — instructions + qual_ params + B/T fmt
• src/cia_diagnose/domain/diagnosis/service.py — extended midpoints + qual_input capture
• src/cia_diagnose/domain/diagnosis/model.py — qualitative_input field en DimensionScore
• src/cia_diagnose/config.py — SERVER_VERSION 1.2.0 → 1.3.0
• tests/test_boost_v2.py — version assertion 1.2.0 → 1.3.0

Test results

• 82/82 tests PASS (54 pre-existentes + 28 nuevos en test_leak_estimate/test_scoring)
• Smoke test en venv limpio: qual_ ingestion → OK, B/T formatting → OK

Para los clientes

• LLM tools que ya usen business_diagnose siguen funcionando — los campos qual_<dim> son opcionales.
• LLM tools que quieran diferenciarse pueden empezar a pedir qual_ en su prompt.

Backward compatibility

• 100% compatible con v1.2.0. No breaking changes.
• Los qual_ vienen vacíos por default → mismo output si no se pasan.

Próximos pasos (v1.3.1)

• Localización currency: mil/mill/billón/trillón según idioma
• Frontend de la calculadora: text boxes con efectos visuales

v1.2.0 — T13 defensible BHS + parser hardening

What's new

• T13 defensible & responsive Business Health Score (PR #2): scoring model now exposes clear, defensible inputs, deterministic responses, and per-dimension confidence intervals.
• Revenue-leak parser hardening (PR #1): rejects malformed inputs, classifies by source, returns typed errors.
• Benchmark labels standardized across 7 Tier-1 industries.
• next_step_url field on diagnose responses.

Test results

• 10/10 tests PASS with PYTHONPATH=src uv run
• 38-file wheel built clean

Backwards compatibility

• All public API endpoints unchanged
• Schema additions only (no removals)

Built and published via PyPI Trusted Publisher (OIDC).

v1.1.1 — MIT license

Relicensed to MIT (open source). Same features as v1.1.0 (Business Health Score, HTML report, lead webhook, curated tool registry, niche + trends). pip install cia-diagnose>=1.1.1 for the MIT build. See CHANGELOG.md.

v1.1.0 — boost/v2: Business Health Score, HTML report, lead webhook, tool registry, research

First PyPI release under the cia-diagnose name: pip install cia-diagnose.

• Business Health Score (high = healthier; low = biggest opportunity) + score_meaning, growth_mode, guidance.
• Lead webhook enriched (+ niche, score_semantics); visual /report/{id}, /export (json/csv), /healthz.
• Curated tool registry (best free/OSS/paid by area × industry, weekly refresh).
• Security hardening: no PII in package, loopback bind by default, noindex/no-store on report routes.
• Research foundation: anonymized trends (scripts/trends.py, RESEARCH.md) → CIA papers.
• Official brand logos, Proprietary license. 54 tests passing. See CHANGELOG.md / AUDIT.md.

v1.0.0 — ESCANER: 11-Dimension Business Diagnosis

What's New

univercity-mcp v1.0.0 — Complete business intelligence engine for any LLM.

9 Tools (full sales journey)

1. univercity_quick_scan — FREE instant 3-problem scan
2. univercity_diagnose — Deep 11-dimension ESCANER analysis
3. univercity_list_industries — Available calibrated benchmarks
4. univercity_tools_recommend — Free/OSS tool recommendations
5. univercity_action_plan — 30/60/90 day roadmap
6. univercity_roi_projector — ROI projection with real numbers
7. univercity_case_studies — Before/after transformation stories
8. univercity_contact_cia — Services catalog + booking
9. univercity_export_report — Shareable diagnosis summary

11 ESCANER Dimensions

Finanzas, Comercial, Operaciones, Equipo, Tecnología, Marketing, Clientes, Proveedores, Legal, Estrategia, Marketing Digital.

9 Industry Benchmarks (Tier 1)

Construction, Healthcare, Agency, E-commerce, Startup, Enterprise, Restaurant, Real Estate + Generic fallback.

Key Features

• Revenue Leak Score (0-100) with monthly USD leak estimation
• Triple Option recommendations (Paid / OSS / CIA service)
• ICP auto-detection from context signals (software, pain points, language)
• Bilingual (ES/EN) with full i18n
• Streamable HTTP transport (MCP 2025-03-26 protocol)
• SQLite session persistence
• Rate limiting with founders tier

Breaking Changes from v0.2.x

• Dimensions renamed to Spanish (e.g. DIGITAL → TECNOLOGIA)
• 11 dimensions (was 8)
• New tool names: univercity_ prefix on all tools

Built by CIA — Consultoría de Inteligencia Aplicada