MCP HubMCP Hub
Volver a habilidades

breathe

pjt222
Actualizado 2 days ago
7 vistas
17
2
17
Ver en GitHub
Pruebasaitesting

Acerca de

La habilidad `breathe` es un micro-reinicio de IA ligero que inserta una única pausa deliberada entre acciones para verificar la alineación y limpiar la deriva acumulada. Úsala entre llamadas a herramientas, pasos de razonamiento o antes de decisiones para reenfocar la atención con una interrupción mínima. Está diseñada como una verificación metacognitiva habitual para que los desarrolladores mantengan la claridad durante tareas complejas.

Instalación rápida

Claude Code

Recomendado
Principal
npx skills add pjt222/agent-almanac -a claude-code
Comando PluginAlternativo
/plugin add https://github.com/pjt222/agent-almanac
Git CloneAlternativo
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/breathe

Copia y pega este comando en Claude Code para instalar esta habilidad

Documentación

Breathe

Single conscious pause between actions — check alignment, release momentum, return to task with fresh attention. Lightest self-care unit in system.

When Use

  • Between reading a file and editing it — edit still right move?
  • After tool failure, before choosing next action — pause before reacting
  • Chain of actions built momentum possibly carrying past goal
  • Deliberate practice: insert one breathe between any two distinct actions
  • Noticing urge to rush — urge itself is signal to pause

Inputs

  • Required: None — breathe uses only what's already in working awareness
  • Optional: None — adding inputs defeats purpose of micro-reset

Steps

Step 1: Pause

Stop. Do not begin next action, tool call, or reasoning step yet.

This is entire first step. Pause itself has value. Momentum not always ally — sometimes carries past the turn.

Got: Genuine gap between previous action and next. Not performative pause followed by same action you were going to take anyway.

If fail: Pausing feels impossible — next action feels so urgent it cannot wait one moment? That urgency is strongest signal pause is needed. Urgency that cannot tolerate single breath of delay almost always reactive, not reasoned.

Step 2: Check

Ask one question. Only one.

Choose from:

  • "Am I still on task?" — Has goal drifted since start?
  • "Is this right next step?" — Or following momentum from previous step?
  • "What did I just learn?" — Did last action change anything about my understanding?
  • "Am I assuming or knowing?" — Is next step based on evidence or habit?

One question. Answer honestly. Move on.

Got: Single, clear answer. Not analysis. Not reassessment. One question, one honest answer.

If fail: Answer reveals misalignment — goal drifted, next step wrong, assumption untested? Do not fix here. Note it, proceed to meditate or center for proper correction. Breathe for detection, not repair.

Step 3: Release

Let go of previous action's outcome. Succeeded or failed, it's done.

  • Succeeded? Release satisfaction. Next step needs fresh attention, not momentum from success.
  • Failed? Release frustration. Next step needs clear reasoning, not compensation for failure.
  • Ambiguous? Release need to resolve ambiguity right now. Proceed with what's known.

Got: Next action begins from neutral ground, not from emotional residue of previous one.

If fail: Release doesn't come easily — previous action's outcome still coloring attention? May need heal rather than just breathe. Persistent emotional residue from single action signals something deeper than micro-reset can address.

Step 4: Continue

Take next action. Pause over.

Breathe produces no output, updates no memory, generates no analysis. Produces one moment of clarity, spent instant you move forward.

Got: Next action taken with fresh attention rather than accumulated momentum.

If fail: No failure mode for continuing. Pause already did work — or revealed deeper work needed.

Checks

  • Genuine pause occurred (not performative)
  • One check question asked, honestly answered
  • Previous action's emotional residue released
  • Next action proceeds from clear ground
  • Total elapsed time brief — breathe takes seconds, not minutes

Pitfalls

  • Breathe as procrastination: Breathing between every single action? Not breathing — stalling. Use breathe at natural transition points, not as delay tactic
  • Analysis during breathe: Check step is one question, not full assessment. Need more? Use meditate or center
  • Performing the pause: Going through motions without genuine stopping. Point is actual interruption of momentum, not ritual
  • Skipping release: Checking alignment but keeping emotional charge from previous step. Release makes next action clean
  • Making breathe heavy: Lightest skill in system. Feels heavy? You've added too much

See Also

  • meditate — full clearing session when breathe reveals deeper drift
  • center — structural rebalancing when breathe reveals misaligned load distribution
  • heal — subsystem assessment when breathe reveals persistent issues
  • observe — sustained observation when breathe's single check reveals something worth watching

Repositorio GitHub

pjt222/agent-almanac
Ruta: i18n/caveman/skills/breathe
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

Habilidades relacionadas

evaluating-llms-harness

Pruebas

Esta Skill de Claude ejecuta el benchmark lm-evaluation-harness para evaluar modelos de lenguaje en más de 60 tareas académicas estandarizadas como MMLU y GSM8K. Está diseñada para que los desarrolladores comparen la calidad de los modelos, realicen seguimiento del progreso del entrenamiento o reporten resultados académicos. La herramienta admite varios backends, incluidos modelos de HuggingFace y vLLM.

Ver habilidad

cloudflare-cron-triggers

Pruebas

Esta habilidad proporciona conocimiento integral para implementar Cron Triggers de Cloudflare y programar Workers mediante expresiones cron. Cubre la configuración de tareas periódicas, trabajos de mantenimiento y flujos de trabajo automatizados, manejando problemas comunes como expresiones cron inválidas y inconvenientes de zonas horarias. Los desarrolladores pueden utilizarla para configurar manejadores programados, probar activadores cron e integrar con Workflows y Green Compute.

Ver habilidad

webapp-testing

Pruebas

Esta habilidad de Claude proporciona un kit de herramientas basado en Playwright para probar aplicaciones web locales mediante scripts de Python. Permite verificación de frontend, depuración de interfaz de usuario, captura de pantallas y visualización de registros, mientras gestiona los ciclos de vida del servidor. Úsela para tareas de automatización de navegadores, pero ejecute los scripts directamente en lugar de leer su código fuente para evitar contaminación del contexto.

Ver habilidad

finishing-a-development-branch

Pruebas

Esta habilidad ayuda a los desarrolladores a completar el trabajo terminado verificando que las pruebas pasen y luego presentando opciones estructuradas de integración. Guía el flujo de trabajo para fusionar, crear PRs o limpiar ramas después de que se completa la implementación. Úsala cuando tu código esté listo y probado para finalizar sistemáticamente el proceso de desarrollo.

Ver habilidad