| Métrica | this 7d | prior 7d | Δ |
|---|---|---|---|
| Activas | 605 | 764 | -20.8% |
| Nuevas | 233 | 332 | -29.8% |
| Link sent | 2 | 1 | +100.0% |
| Calls booked | 0 | 0 | — |
| Booked/Link % | 0.0% | 0.0% | +0.0pp |
| Booked/Activas % | 0.0% | 0.0% | +0.0pp |
| Booked/Nuevas % | 0.0% | 0.0% | +0.0pp |
| Revenue variable | $0.00 | $0.00 | — |
| Período | Activas | Nuevas | Link sent | Calls booked | Booked/Link | Booked/Activas | Booked/Nuevas |
|---|---|---|---|---|---|---|---|
| Semana actual (7d) | 605 | 233 | 2 | 0 | 0.0% | 0.0% | 0.0% |
| Semana anterior (7d) | 764 | 332 | 1 | 0 | 0.0% | 0.0% | 0.0% |
| Día | Activas | Nuevas | Link sent | Calls booked | Booked/Link | Booked/Activas | Booked/Nuevas |
|---|---|---|---|---|---|---|---|
| 09-27 | 14 | 9 | 0 | 0 | — | 0.0% | 0.0% |
| 09-26 | 13 | 9 | 0 | 0 | — | 0.0% | 0.0% |
| 09-25 | 76 | 15 | 0 | 0 | — | 0.0% | 0.0% |
| 09-24 | 90 | 33 | 0 | 0 | — | 0.0% | 0.0% |
| 09-23 | 112 | 55 | 0 | 0 | — | 0.0% | 0.0% |
| 09-22 | 115 | 47 | 2 | 0 | 0.0% | 0.0% | 0.0% |
| 09-21 | 185 | 65 | 0 | 0 | — | 0.0% | 0.0% |
| 09-20 | 58 | 50 | 0 | 0 | — | 0.0% | 0.0% |
| 09-19 | 30 | 19 | 0 | 0 | — | 0.0% | 0.0% |
| 09-18 | 96 | 37 | 0 | 0 | — | 0.0% | 0.0% |
| 09-17 | 114 | 63 | 1 | 0 | 0.0% | 0.0% | 0.0% |
| 09-16 | 116 | 56 | 0 | 0 | — | 0.0% | 0.0% |
| 09-15 | 103 | 41 | 0 | 0 | — | 0.0% | 0.0% |
| 09-14 | 247 | 66 | 0 | 0 | — | 0.0% | 0.0% |
| BP | Sev | Estado | Detalle |
|---|---|---|---|
| BP-1 | HIGH | PASS | Template compliance The agent declares template 'membership-sales' in meta.md, but no matching template file exists in the repository. However, system.md contains all required sections for a principles-based agent: Identity (Identidad), Voice (Voz), Flow (Flujo), Calibration (Calibración), Principles (Principios), Tools (Herramientas), and NO_RESPONSE behaviors. The content structure and intent align with the principles template specification. |
| BP-2 | MEDIUM | PASS | Example coverage and cap (7 base scenarios + conditional, ≤20 total) examples.md contains 20 examples covering all 7 base scenarios (happy path, resource/link delivery, objection handling, graceful disqualification, emotional lead, re-engagement, off-topic) plus two conditional scenarios (keyword trigger for JCR/Jockey Club in Example 7, price objection in Example 5). Total count is exactly 20, meeting the cap requirement. |
| BP-3 | HIGH | PASS | Examples align with the flow All 20 examples in examples.md follow the flow steps defined in system.md (Conocer → Motivación → Sede → Valor → Precios → Objeciones → Pago/Datos → Cierre) in order, without skipping required steps. Each example reaches a terminal state: link sent, fallback resource, NO_RESPONSE, or objection handled with CTA. |
| BP-4 | CRITICAL | PASS | No internal contradictions in system.md system.md contains no self-contradictions. All stated rules are either defaults with explicitly named exceptions (e.g., 'Tarjeta: link directo... Única excepción, débito automático: ahí SÍ los datos van primero') or apply to different preconditions (different steps, different lead types). The flow and calibration sections coexist without conflicting outputs under any shared precondition. |
| BP-5 | CRITICAL | PASS | No cross-file contradictions All SDK files (system.md, examples.md, objections.md, keywords.json, program.json, resources.json, knowledge_base.md, personal_story.md) agree on behavior, parameters, and data. Examples demonstrate rules from system.md operating correctly (e.g., Example 7 shows JCR flow with cuota local $41.040 matching knowledge_base.md; Example 8 shows débito automático with data-first matching system.md Step 7). No file declares behavior or data that contradicts another file under the same precondition. |
| BP-6 | MEDIUM | PASS | Voice consistency across examples All 20 examples in examples.md use the same persona (Sofi, casual, energetic, rioplatense), vocabulary (voseo, informal tone), and formality level. No jarring tonal shifts; a reader could believe all examples are the same agent talking to different leads. |
| BP-7 | CRITICAL | PASS | Unguarded actions (PAT-004) All risky actions in system.md's flow include inline guard clauses. Sending links is guarded by 'Confirmá la compra antes de mandar link o CBU' (Step 7). Sending CBU is guarded by the same. Qualifying/booking is guarded by lead confirmation. The 'También tenemos Estación Club Dúo' mention (previously dismissed as passive informational add-on) is not a targeted offer requiring a guard. Multiple worked examples (1, 7, 8, 14, 15) demonstrate guards firing correctly. |
| BP-8 | HIGH | PASS | No stale version language in examples All closing/CTA language in examples.md matches the current flow definition in system.md. Examples use current phrasing ('ya te cargué', 'te esperamos', 'éxitos', 'NO_RESPONSE') consistent with the Flujo and NO_RESPONSE sections. No stale or outdated CTAs detected. |
| BP-9 | MEDIUM | PASS | No duplicated instructions Each behavioral instruction appears in exactly one canonical location. For example, the rule 'Pedir nombre Y apellido por separado, nunca nombre completo' is stated once in system.md Step 7 and cross-referenced in program.json data_collection (not duplicated). The rule 'Confirmá la compra antes de mandar link' appears once in system.md Step 7 and is demonstrated in examples (not duplicated). No instruction is restated in two locations. |
| BP-10 | MEDIUM | PASS | Rule-vs-principle balance Behavioral guidance (tone, flow, conversational style) is expressed via Voz section + 11 worked examples showing the voice in action, and via Principios section. Hard rules remain for factual constraints (prices, phone numbers, CBU, sedes) and safety boundaries (no razonamientos internos, no inventar). The balance is appropriate: principles + examples govern behavior; rules govern facts. |
| Check | Sec | Estado | Notas |
|---|---|---|---|
| 6a | 1.8 | PASS | NO_RESPONSE present (limits not checked, --no-db) |
| 6c | 3.x | PASS | 141 lines |
| 6d | 1.2 | PASS | 20 examples |
| 6e | 1.3 | PASS | 7 principles |
| 6g | PAT-006 | FAIL | keywords defined but no handling in system.md |
| 6h | Anti-pat | PASS | no 'burbuja' |
| 6i | Anti-pat | PASS | no HTML comments |
| 6j | Anti-pat | FAIL | 68 em-dashes across 1 file(s); first: system.md:3 |
| 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 | WARN | 2 keywords.json entr(ies) not named in system.md (no recent operator ask): ['amigo', 'personal trainer'] — likely legacy, review or remove |
| 6s | Hygiene | WARN | 3 expired date reference(s) in SDK files |
| 6t | Coverage | WARN | 3/7 principle(s) without example: #3 El nombre es solo lo que el lead responde a "cómo te llamás?"., #4 Confirmá la compra antes de mandar link o CBU., #7 `NO_RESPONSE` es tu único stop.; 3/7 flow step(s) without example: #2 Motivación, #4 Valor, #6 Objeciones |
| Fecha | Tema | Tags |
|---|---|---|
| 2026-09-28 | critical data gap trial to membership | action-item, measurement-gap, conversion-tracking |
| 2026-09-28 | weekly funnel report sep 21 28 | metrics, funnel-analysis, conversion, keyword-performance |
| 2026-09-24 | super sur promo first month 50 percent | pricing, promotion, sede-sur, debito-automatico, critical-update |
| 2026-09-24 | super sur promo implementation | pricing, promotion, sede-sur, debito-automatico, critical |
| 2026-09-21 | weekly funnel report sep 14 20 | metrics, funnel, performance, action-items |
| 2026-09-21 | weekly funnel and conversion report sep14 20 | metrics, funnel, conversion, content-strategy, followup |
| 2026-09-18 | manychat instagram routing configuration | platform-config, instagram, manychat, integration |
| 2026-09-16 | manychat instagram threading conflict | platform-bug, instagram, manychat, workflow-blocker |
| 2026-09-16 | viumi pricing config error fix v114 | pricing, bug, viumi, knowledge-base, deployment |
| 2026-09-16 | viumi pricing ambiguity fix deployed | pricing, viumi, bug-fix, deployment |
| 2026-09-08 | viumi information audit | audit, knowledge-base |
| 2026-09-07 | weekly traffic analysis sep1 7 | analytics, traffic, growth |
| 2026-09-03 | comprehensive pricing inconsistency audit | audit, pricing, critical-inconsistencies |
| 2026-09-03 | permanence inconsistency da 4 vs 6 months | inconsistency, audit, permanence-policy |
| 2026-09-03 | da sur price update 49999 to 55999 | pricing, deployment, critical-update |
| 2026-09-02 | sofi south sede offering summary sept 2 | offer-structure, sede-sur, sofi-agent |
| 2026-09-02 | deployment protocol cross field verification | process, protocol, deployment-hygiene |
| 2026-09-02 | promo 24999 still offered 8am due to kb lag | bug-analysis, prompt-vs-knowledge-base, da-sur-pricing |
| 2026-09-02 | old price 66000 da multisede residual | bug, pricing, examples-field, da-multisede |
| 2026-09-01 | da sur price correction 24999 to 49999 | pricing, deployment, urgent |
| ID | Estado | Cambio pedido |
|---|---|---|
| ⚠️ | ⚠️ | Parcialmente cubierto o depende de plataforma |
| 23 | pendiente | Bug duplicación de mensajes |
| 63 | pendiente | Agente no responde mensajes de primer contacto cortos ("Hola", "consulta") — evaluator posiblemente demasiado restrictivo |
| 64 | pendiente | Doble saludo: fix = First Contact Message a nivel plataforma (2 mensajes fijos) |
| 69 | pendiente | Follow-ups automáticos: 3 workflows (lo-pienso 23h, gracias-muerta 23h, post-precios 23h) |
| X2 | pendiente | Doble saludo (fraccionamiento) |
| Y1 | pendiente | Conv muerta ante "gracias" suelto |
| Y2 | pendiente | Follow-ups pendientes (Julian pedía desde 17/03) |
| Señal | Valor | Evidencia (cita del cliente) |
|---|---|---|
| Enojado con la plataforma | sí | — |
| Es queja | sí | — |
| Queja constructiva (accionable) | sí | — |
| Señal | Cita textual | Fecha |
|---|---|---|
| Bug recurrente reportado por el cliente: IA se queda en loop o se corta sin completar info — preocupación principal post-launch | “None” | — |
| Cliente reporta que el agente repregunta lo mismo de la misma manera, sin variación — señal de fricción operativa sostenida | “None” | — |
| Cliente (Julian/Santiago) manda feedback directo por Slack y exige aplicación inmediata — múltiples versiones en un día; alta demanda operativa sobre Ninjo | “None” | — |
| Engagement activo: cliente provee actualizaciones de pricing con reglas complejas y detalladas, indicando inversión en el producto | “None” | 2026-09-24 |
| Cliente exige protocolo de auditoría cruzada obligatoria antes de cada deploy — reacción a errores previos de la plataforma que afectaron precios comunicados | “None” | 2026-09-03 |
Un creador que pide cambios concretos está enganchado (mejor señal que el silencio).
| ID | Estado | Cambio pedido |
|---|---|---|
| ⚠️ | ⚠️ | Parcialmente cubierto o depende de plataforma |
| 23 | pendiente | Bug duplicación de mensajes |
| 63 | pendiente | Agente no responde mensajes de primer contacto cortos ("Hola", "consulta") — evaluator posiblemente demasiado restrictivo |
| 64 | pendiente | Doble saludo: fix = First Contact Message a nivel plataforma (2 mensajes fijos) |
| 69 | pendiente | Follow-ups automáticos: 3 workflows (lo-pienso 23h, gracias-muerta 23h, post-precios 23h) |
| X2 | pendiente | Doble saludo (fraccionamiento) |
| Y1 | pendiente | Conv muerta ante "gracias" suelto |
| Y2 | pendiente | Follow-ups pendientes (Julian pedía desde 17/03) |