Review — gaston.castriota

← overview

ventana últimos 7d vs 7d previos · tz America/Argentina/Buenos_Aires · último cambio Ninjo: 2026-05-21 14:39 (deploy config, hace 13d)

Nuevas (7d)

prior 2

Calls booked

— de link

Link sent

0.0% de activas

Revenue variable

83 msgs

Resumen — período vs período

Métrica	this 7d	prior 7d	Δ
Activas	40	21	+90.5%
Nuevas	10	2	+400.0%
Link sent	0	0	—
Calls booked	0	0	—
Booked/Link %	—	—	—
Booked/Activas %	0.0%	0.0%	+0.0pp
Booked/Nuevas %	0.0%	0.0%	+0.0pp
Revenue variable	$2.91	$1.89	+54.0%

Funnel — semana actual vs anterior

Período	Activas	Nuevas	Link sent	Calls booked	Booked/Link	Booked/Activas	Booked/Nuevas
Semana actual (7d)	40	10	0	0	—	0.0%	0.0%
Semana anterior (7d)	21	2	0	0	—	0.0%	0.0%

Ver desglose por día

Día	Activas	Nuevas	Booked/Link	Booked/Activas	Booked/Nuevas
06-02	2	0	—	0.0%	—
06-01	8	1	—	0.0%	0.0%
05-31	5	2	—	0.0%	0.0%
05-30	10	2	—	0.0%	0.0%
05-29	5	2	—	0.0%	0.0%
05-28	7	3	—	0.0%	0.0%
05-27	3	0	—	0.0%	—
05-26	5	2	—	0.0%	0.0%
05-25	5	0	—	0.0%	—
05-24	3	0	—	0.0%	—
05-23	0	0	—	—	—
05-22	1	0	—	0.0%	—
05-21	4	0	—	0.0%	—
05-20	3	0	—	0.0%	—

run 2026-06-02T17:28:49+00:00 · fuente: live DB state (daily_audit pipeline)

10/10

BP semánticos PASS

top-10 best practices

14/16

Determinísticos PASS

0 FAIL · 2 WARN

C0 · H0 · M0

Fails por severidad

critical / high / medium

—

Drift

DB vs repo local

BP semánticos (top-10 best practices)

BP-1

BP-2

BP-3

BP-4

BP-5

BP-6

BP-7

BP-8

BP-9

BP-10

BP	Sev	Estado	Detalle
BP-1	HIGH	PASS	Template compliance All 7 required sections for principles template are present in system.md: Identity, Voice, Flow, Calibration, Principles, Tools, and NO_RESPONSE. Each section matches the template's specification and intent.
BP-2	MEDIUM	PASS	Example coverage and cap (7 base scenarios + conditional, ≤20 total) examples.md contains 19 examples covering all 7 base scenarios (happy path, resource delivery, objection handling, disqualification, emotional lead, re-engagement, off-topic) plus keyword-trigger examples. Total count is 19, within the ≤20 cap. evidence (1) {'file': 'examples.md', 'quote': '## Ejemplo NIVEL 1: Dolor puntual + baja conciencia → masterclass [through] ## Ejemplo 19: Consecuencia temporal anclada antes del compromiso + urgencia en el link'}
BP-3	HIGH	PASS	Examples align with the flow Every example in examples.md follows the flow steps defined in system.md (Apertura → Explorar → Profundizar → Diagnóstico → Cierre/NO_RESPONSE) in order, and each reaches a terminal state (link sent, NO_RESPONSE, or graceful close).
BP-4	CRITICAL	PASS	No internal contradictions in system.md system.md is internally consistent. The Calibration table and Flow section work together without contradiction. The 'no sé qué hacer' case is explicitly documented as an exception in the Profundizar step with a named action (labeling + advance to Diagnóstico), not a contradiction.
BP-5	CRITICAL	PASS	No cross-file contradictions All SDK files are consistent. Keywords.json delivers the unifying greeting; system.md governs the agent's generated responses. Examples.md demonstrates the flow correctly. resources.json URLs match those referenced in system.md and examples.md. program.json pricing rules align with objections.md and system.md (never mention price in DMs).
BP-6	MEDIUM	PASS	Voice consistency across examples All examples use consistent voice: direct, calm, rioplatense (voseo), short messages, labeling before confrontation, no opening ¿ or ¡. Tone and vocabulary remain uniform across all 19 examples.
BP-7	CRITICAL	PASS	Unguarded actions (PAT-004) All risky actions in system.md have inline guards. get_resource calls are guarded by Calibration conditions (dolor concreto, patrón propio, etc.). Calendly link is guarded by the 4-signal NIVEL 4 condition. NO_RESPONSE is guarded by explicit triggers (post-link without objection, emoji-only, off-topic, etc.). Examples demonstrate guards firing correctly.
BP-8	HIGH	PASS	No stale version language in examples All examples use current flow language and CTAs matching system.md. No outdated phrasing or deprecated flow steps appear in examples.md.
BP-9	MEDIUM	PASS	No duplicated instructions Each behavioral instruction has a single canonical home. The 'no sé qué hacer' rule appears once in Profundizar step; the labeling principle appears once in Voz section; recalibration logic appears once in Calibration table. No material duplication across sections.
BP-10	MEDIUM	PASS	Rule-vs-principle balance Behavioral guidance (tone, labeling, message length, rhythm adaptation) is expressed via Voice section + examples. Hard rules are reserved for factual constraints (pricing approach, eligibility, NO_RESPONSE triggers). The balance favors principles and examples over prescriptive rules.

Checks determinísticos

Check	Sec	Estado	Notas
6a	1.8	PASS	NO_RESPONSE present (limits not checked, --no-db)
6c	3.x	PASS	129 lines
6d	1.2	PASS	18 examples
6e	1.3	PASS	7 principles
6g	PAT-006	WARN	non-STEP-0 keyword handling
6h	Anti-pat	PASS	no 'burbuja'
6i	Anti-pat	PASS	no HTML comments
6j	Anti-pat	PASS	no em-dashes
6k	PAT-006	PASS	no tool syntax leak
6l	Resources	PASS	no placeholder URLs
6m	Hygiene	PASS	no audit metadata in prompt files
6n	PAT-008	PASS	URL forms consistent across files
6q	PAT-001	PASS	no instruction duplication into examples.md, keywords.json, program.json, resources.json
6r	Keyword-drift	PASS	all 6 keyword(s) consistent across keywords.json + system.md (no DB)
6s	Hygiene	PASS	no expired dates in live SDK
6t	Coverage	WARN	7/7 principle(s) without example: #1 Liderás, no contenés., #2 Diagnosticá antes de ofrecer., #3 Entrega inmediata., #4 No contenido gratuito., #5 No precio en DMs.; 4/5 flow step(s) without example: #1 Apertura., #2 Explorar., #3 Profundizar., #4 Diagnóstico y entrega.

canal — · sync 2026-06-02

Digests

6 en últimos 14d

Cambios pedidos / pendientes

del checklist

Threads

en el canal

2026-06-02

Última sync

del digest

Cosas nuevas (digests recientes)

Fecha	Tema	Tags
2026-06-01	weekly activation report lead volume systemization	metrics, lead-generation, cta-strategy, action-item
2026-06-01	weekly activation report volume recovery	reporting, volume, activation, cta-strategy
2026-05-25	weekly volume report may 18 24	metrics, weekly-report, volume-alert
2026-05-25	weekly funnel report may 18 24	metrics, funnel, volume-alert
2026-05-20	biblioteca premium 5 conversation examples	ejemplos-conversacion, nivel-3, biblioteca-premium, escalera-valor
2026-05-20	prompt simplification and logic conflict resolution	prompt-refinement, conflicting-instructions, escalera-logica, deployment-pending
2026-05-19	escalera de valor 4 niveles deployment	escalera-de-valor, product-strategy, deployment, diagnostic-framework
2026-05-19	escalera de valor 4 levels deployed	escalera-de-valor, routing-logic, v73, v74, structural-change
2026-05-19	no free content mandate strict policy	free-content-ban, policy-enforcement, resource-delivery, anti-freebie
2026-05-19	escalera program update v79	v79, escalera-de-valor, program-update, examples-masterclass, nivel-1-fix, diagnose-before-offer
2026-05-19	masterclass freebie removed v77	masterclass, v77, resources-cleanup, escalera-clarification, paid-masterclass-link
2026-05-19	no free content rule v75	no-free-content, v75, principio-14, hard-rule
2026-05-19	resources json update technical challenge	technical-issue, resources-config, api-gateway
2026-05-18	keywords only cold entry deep convos natural	prompt-iteration, conversion-logic, product-routing
2026-05-18	eliminate keyword triggers in deep conversations	conversation-flow, keyword-strategy, naturalism
2026-05-18	immediate product delivery no delay language	authority, conversion, prompt-iteration
2026-05-18	immediate delivery principle no delays	authority, delivery, trust-erosion
2026-05-18	weekly metrics ay 11 17 volume drop	metrics, monitoring
2026-05-18	product first hierarchy call exception only	conversion-hierarchy, routing-logic, critical-structural-change
2026-05-17	handle product objections by recalibrating	objection-handling, product-scaling, emotional-calibration

Cambios pedidos por el creador / pendientes

ID	Estado	Cambio pedido
X4	pendiente	Flujo de calificación no es bloqueante: el agente avanza etapas sin validar condiciones. Pato deployó cambios 22:35 UTC 2026-04-22 — pendiente confirmación de Gastón

sentiment del cliente → Ninjo · caché (2026-06-02 19:34:59) · modelo claude-sonnet-4-6 · ventana 7d

Sentiment score

neutral_con_señales_de_engagement_activo

Bajo

Churn risk

del lado del cliente

Cambios pedidos

pendientes en checklist

Volumen cliente

agente: 10 convs

Matriz de volumen 🚨

alarma real

Cliente callado mientras el agente tiene tráfico: desenganche / churn silencioso. Abrir loop proactivo.

Sentiment del cliente

📋 Sin citas textuales del cliente en los digests disponibles (todos los resúmenes son voz de Ninjo/Cortex), no es posible asignar sentimiento negativo ni positivo con evidencia directa. El patrón operativo muestra un cliente altamente comprometido (44 threads, múltiples aprobaciones de versiones, iteraciones semanales), sin señales de churn. Riesgo: la fricción recurrente con la calificación de leads (2+ rondas sin cierre definitivo) podría escalar a frustración si no se resuelve. Acción recomendada: el CS debe solicitar al cliente una valoración directa del estado del agente y documentar sus palabras textuales para poder medir sentimiento real en el próximo ciclo.

Señal	Valor	Evidencia (cita del cliente)
Enojado con la plataforma	no	—
Es queja	no	—
Queja constructiva (accionable)	no	—

Evidencia citada

Señal	Cita textual	Fecha
El cliente (Gastón/Yamila) aprueba iteraciones de prompt de forma activa y recurrente, señal de engagement sostenido con la plataforma	“None”	2026-05-21
Yamila aprueba explícitamente revisión quirúrgica de prompt — participación directa del equipo en la evolución del agente	“None”	2026-05-11
Gastón reescribe copy de primer mensaje cuando no genera la emoción/profundidad que busca — exigencia alta pero no dirigida contra Ninjo	“None”	—
Problema de leads no calificadas identificado como tema recurrente #1 con dos rondas de overhaul — fricción operativa repetida pero tratada colaborativamente	“None”	—

Cambios pedidos por el creador (Métrica 1)

Un creador que pide cambios concretos está enganchado (mejor señal que el silencio).

ID	Estado	Cambio pedido
X4	pendiente	Flujo de calificación no es bloqueante: el agente avanza etapas sin validar condiciones. Pato deployó cambios 22:35 UTC 2026-04-22 — pendiente confirmación de Gastón