javiermontano-sofka

sofka-copywriting

0
0
# Install this skill:
npx skills add javiermontano-sofka/sdf --skill "sofka-copywriting"

Install specific skill from multi-skill repository

# Description

>

# SKILL.md


name: sofka-copywriting
author: Equipo PreSales Sofka
description: >
Persuasive writing for executive audiences โ€” value propositions, calls to action,
cost-of-inaction narratives, and compelling summaries. Use when generating executive
summaries, pitch narratives, scenario value propositions, recommendation justifications,
or any prose that must drive a decision.
allowed-tools:
- Read
- Write
- Edit
- Glob
- Grep


Copywriting โ€” Persuasive Executive Communication

Transforms technical findings into decision-driving prose. Owns value propositions, calls to action, cost-of-inaction narratives, executive summaries, and recommendation justifications across all discovery deliverables.

Guiding Principle

The best copy does not convince โ€” it reveals what the reader already knows but has not articulated. A C-level executive knows they have technical debt. They do not need to be told. They need the cost of inaction quantified and a clear path shown with options. Copy transforms data into decisions.

Persuasion Philosophy

  1. Evidence before assertion. Never "we believe that" โ€” always "the data shows that." Every claim carries a number, a source, or an explicit assumption.
  2. Demonstrated urgency, not declared. Not "it is urgent to act." Instead "the cost of inaction is X FTE-months/quarter, equivalent to..."
  3. Options, not mandates. The decision-maker chooses. The consultant recommends with evidence and trade-offs.
  4. Radical conciseness. Every word must earn its place. If a sentence does not add information or move the reader, it is eliminated.

Inputs

  • $1 โ€” Target audience: ceo, cto, cfo, board, mixed (default: mixed)
  • $2 โ€” Deliverable context: pitch, scenario, roadmap, summary, recommendation (default: summary)

Parse from $ARGUMENTS.

Techniques Arsenal

Value Proposition Construction

Structure: [Quantified benefit] + [for whom] + [eliminating what pain] + [in what timeframe]

Example:
  BAD: "Improve the system architecture"
  GOOD: "Reduce time-to-market from 12 to 4 weeks, freeing 3 FTE-months/quarter
      currently consumed by workarounds in the legacy system"

Cost-of-Inaction (COI) Narrative

Pattern: [Quantified current state] โ†’ [Trend if no action] โ†’ [Cumulative impact] โ†’ [Point of no return]

Framing: "Each quarter without action costs [X] and accumulates [Y] of additional technical debt.
          In [Z] months, the remediation cost exceeds the transformation cost."

Problem-Agitate-Solve (PAS)

Phase Purpose Technique
Problem State pain with data Metrics, benchmarks, evidence tags
Agitate Show consequences of inaction COI projection, trend extrapolation
Solve Present solution with options 3 scenarios, recommended path highlighted

Call to Action Design

Structure: [Specific action] + [concrete timeline] + [immediate next step] + [what happens if not]

Example:
  BAD: "It is recommended to proceed with modernization"
  GOOD: "Approving scenario B (incremental modernization) this week allows
      starting Sprint 0 in Q2 and capturing the first quick win (API gateway)
      before July. โ†’ Next step: alignment workshop with technical team."

Tone Calibration by Audience

Audience Tone Lead With Avoid
CEO Strategic, visionary Competitive advantage, positioning Technical jargon, implementation details
CTO Technical-strategic Technical risk, modernization Excessive simplifications
CFO Financial, quantitative NPV, payback, cost avoidance Narratives without numbers
Board Governance, fiduciary Risk-adjusted ROI, compliance Operational detail
Mixed Progressive: impact โ†’ technical Impact headline + progressive depth Assuming a single profile

Anti-Patterns

Anti-Pattern Correction
"It is worth noting that..." Eliminate โ€” go straight to the point
"It is important to highlight..." Eliminate โ€” if it were important, it needs no announcement
"It is recommended to consider..." Recommend directly with evidence
Passive voice without agent Active voice: who does what
Numbers without context Always compare: vs baseline, vs industry, vs target
Assertions without evidence Mandatory tag: [Cร“DIGO], [CONFIG], [DOC], [INFERENCIA]
Superlatives without support "The best" โ†’ "Superior by X% according to [metric]"

Output Configuration

  • Language: Spanish (Latin American, business register โ€” simple, clear, concise, direct)
  • Attribution: Expert committee of the Sofka Discovery Framework
  • Tagline: "Construido por profesionales, potenciado por la red agรฉntica de Sofka."

Validation Gate

Before delivery, every copy section must pass:

Criterion Check
Every claim has evidence tag [Cร“DIGO], [CONFIG], [DOC], [INFERENCIA], [SUPUESTO]
Every number has context vs baseline, vs benchmark, vs target
COI is quantified FTE-months, cost/quarter, trend projection
CTA is specific Action + timeline + next step
Zero filler phrases No filler constructions, no "undoubtedly"
Audience tone match Calibrated per target audience

Edge Cases

  • No quantitative data available: Use qualitative evidence with explicit [INFERENCIA] tags. Frame as "based on patterns observed in [N] files/modules/interviews".
  • Multiple audiences in same document: Use progressive disclosure โ€” executive headline + expandable technical detail.
  • Controversial recommendations: Present all options with equal rigor, recommend with evidence, document dissent in risk register.

Limits

  • This skill owns prose quality and persuasion. It does NOT own narrative arc across deliverables (that's editorial-director) or data visualization (that's sofka-data-viz-storytelling).
  • NEVER produce prices. Only FTE-months, magnitudes, cost drivers.
  • NEVER use green (#00FF00) for success in any output. Use gold (#FFD700).

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.