/ Directorio / Playground / strategic-partner
← Todos los servidores ↗ GitHub
● Comunidad JimmySadek ⚡ Instantáneo

strategic-partner

por JimmySadek · JimmySadek/strategic-partner

Chief-of-Staff skill for Claude Code — separates deciding from building. Challenges assumptions, forces Path A/B/C, gates before implementation.

JimmySadek/strategic-partner makes Claude act as an advisory persona rather than an executor. Two-session loop: an Advisory session (persistent) interrogates assumptions, presents Path A/B/C with tradeoffs, routes tasks, and keeps decision history — while Execution sessions (ephemeral) receive self-contained prompts and build with clean context. Key gates: premise challenge (4 trigger conditions), forced alternatives (3 paths), confidence labels ([SAFE] / [RISK]), Advisory Completion Gate (5 conditions), 14 named cognitive patterns. Core principle: 'The SP never builds. The executor never decides.'

▶ Ver demo 📦 Instalar 💡 Casos de uso

Por qué usarlo

Características clave

Demo en vivo

Cómo se ve en la práctica

strategic-partner-skill.replay ▶ listo
0/0

Instalación

Repositorio del autor · Documentación oficial de Skills de Claude

Casos de uso

Usos del mundo real: strategic-partner

Run a decision through the advisor before touching code

👤 Tech leads and founders who suspect they'd regret building too soon ⏱ ~30 min intermediate

Cuándo usarlo: You're about to ask Claude Code to build something substantial. Pause and consult the advisor first.

Flujo
  1. Open the advisory session
    Using strategic-partner — I'm about to add a queue system to handle email sends. Challenge my premise first.✓ Copiado
    → Claude probes whether a queue is actually needed vs simpler alternatives
  2. Receive Path A/B/C
    Give me three paths with tradeoffs.✓ Copiado
    → Three distinct paths, each labeled [SAFE] or [RISK], with honest downsides
  3. Cross Completion Gate
    I pick Path B. Prepare the executor prompt.✓ Copiado
    → Gate check runs; if conditions aren't met, you're pushed back to advisory
  4. Hand off to executor
    Open a fresh Claude Code session and paste the executor prompt.✓ Copiado
    → Execution session works with clean context

Resultado: Work that's been decided deliberately rather than drifted into.

Errores comunes
  • Using the advisor as a rubber stamp — If you're always picking 'Path A — do what I said', the premise challenge isn't working — provide real context
Combinar con: github

Run a weekly strategy review with the persistent advisor

👤 Solo operators without a real chief-of-staff ⏱ ~30 min beginner

Cuándo usarlo: Friday afternoon, closing out the week.

Flujo
  1. Review decisions made this week
    strategic-partner — review this week's advisory log. Which decisions do I now have evidence to revise?✓ Copiado
    → Specific call-outs of decisions where execution revealed something new
  2. Plan next week
    What's the one decision next week that most deserves premise challenge?✓ Copiado
    → Pinpoint rather than a to-do list

Resultado: A weekly audit that compounds across months.

Combinar con: planning-with-files

Combinaciones

Combínalo con otros MCPs para multiplicar por 10

strategic-partner-skill + planning-with-files

Persist the advisor's decision log across sessions in findings.md / progress.md

Use planning-with-files — keep decisions.md updated every advisory session.✓ Copiado
strategic-partner-skill + github

When executor session commits, reference the decision from advisor log in the PR body

After executor finishes, link the advisor decision ID in the PR description.✓ Copiado

Herramientas

Lo que expone este MCP

HerramientaEntradasCuándo llamarCoste
premise_challenge intended work + context Before any substantial build 0
forced_alternatives problem statement Once premise passes; before committing to one path 0
completion_gate chosen path + context Before crafting executor prompt 0
executor_prompt_craft decision + context Handoff to execution 0
decision_log current session state Throughout advisory session 0

Coste y límites

Lo que cuesta ejecutarlo

Cuota de API
None
Tokens por llamada
Advisor sessions carry persistent context — can grow
Monetario
Free — skills are local files
Consejo
Trim the decision log periodically; stale decisions bloat context.

Seguridad

Permisos, secretos, alcance

Almacenamiento de credenciales: No credentials — pure prompts
Salida de datos: None

Resolución de problemas

Errores comunes y soluciones

Wrong trigger keywords — no premise challenge happens

Phrase as 'challenge my premise' or 'advisor mode' explicitly; 'help me build X' skips advisory.

Alternativas

strategic-partner vs otros

AlternativaCuándo usarlaContrapartida
planning-with-filesYou want disk-based progress memory rather than decision gatingComplementary — different purposes
Carmack CouncilYou want multi-expert review rather than chief-of-staff personaDifferent phase — Council reviews, SP decides upstream

Más

Recursos

📖 Lee el README oficial en GitHub

🐙 Ver issues abiertas

🔍 Ver todos los 400+ servidores MCP y Skills