| Métrica | this 7d | prior 7d | Δ |
|---|---|---|---|
| Activas | 1376 | 111 | +1139.6% |
| Nuevas | 735 | 44 | +1570.5% |
| Link sent | 102 | 7 | +1357.1% |
| Calls booked | 41 | 4 | +925.0% |
| Booked/Link % | 40.2% | 57.1% | -16.9pp |
| Booked/Activas % | 3.0% | 3.6% | -0.6pp |
| Booked/Nuevas % | 5.6% | 9.1% | -3.5pp |
| Revenue variable | $96.78 | $2.55 | +3695.3% |
| Período | Activas | Nuevas | Link sent | Calls booked | Booked/Link | Booked/Activas | Booked/Nuevas |
|---|---|---|---|---|---|---|---|
| Semana actual (7d) | 1376 | 735 | 102 | 41 | 40.2% | 3.0% | 5.6% |
| Semana anterior (7d) | 111 | 44 | 7 | 4 | 57.1% | 3.6% | 9.1% |
| Día | Activas | Nuevas | Link sent | Calls booked | Booked/Link | Booked/Activas | Booked/Nuevas |
|---|---|---|---|---|---|---|---|
| 06-02 | 165 | 107 | 9 | 3 | 33.3% | 1.8% | 2.8% |
| 06-01 | 170 | 113 | 7 | 4 | 57.1% | 2.4% | 3.5% |
| 05-31 | 160 | 112 | 9 | 1 | 11.1% | 0.6% | 0.9% |
| 05-30 | 273 | 104 | 22 | 11 | 50.0% | 4.0% | 10.6% |
| 05-29 | 230 | 111 | 21 | 8 | 38.1% | 3.5% | 7.2% |
| 05-28 | 277 | 124 | 20 | 6 | 30.0% | 2.2% | 4.8% |
| 05-27 | 101 | 64 | 14 | 8 | 57.1% | 7.9% | 12.5% |
| 05-26 | 20 | 10 | 2 | 1 | 50.0% | 5.0% | 10.0% |
| 05-25 | 17 | 5 | 4 | 2 | 50.0% | 11.8% | 40.0% |
| 05-24 | 9 | 4 | 0 | 0 | — | 0.0% | 0.0% |
| 05-23 | 17 | 2 | 0 | 0 | — | 0.0% | 0.0% |
| 05-22 | 12 | 3 | 0 | 0 | — | 0.0% | 0.0% |
| 05-21 | 17 | 11 | 1 | 1 | 100.0% | 5.9% | 9.1% |
| 05-20 | 19 | 9 | 0 | 0 | — | 0.0% | 0.0% |
| BP | Sev | Estado | Detalle |
|---|---|---|---|
| BP-1 | HIGH | N/A | Template compliance Agent declares template 'universal_dm (custom v5Config)' but no matching template file exists in the rubric's template library. The rubric specifies BP-1 applies to agents matching a known template (e.g., 'principles'). Without a canonical template definition to compare against, compliance cannot be verified. evidence (2)
|
| BP-2 | MEDIUM | PASS | Example coverage and cap (7 base scenarios + conditional, ≤20 total) examples.md contains 14 examples covering 6 of 7 base scenarios (happy path, resource delivery, objection handling, disqualification, re-engagement, off-topic). Keyword-trigger examples are present (Examples 1, 2, 3, 5 show keyword-driven flows). Price-objection example is present (Example 7). Total count is 14, well under the 20-example cap. |
| BP-3 | HIGH | PASS | Examples align with the flow All 14 examples follow the documented flow (Etapas 1–7) in order and reach terminal states (booking confirmed, NO_RESPONSE, or objection handled). No example ends mid-flow or skips required steps. |
| BP-4 | CRITICAL | PASS | No internal contradictions in system.md system.md contains one potential internal tension: Etapa 3 states 'Si ya describió sus resultados... no repetir la pregunta', yet Etapa 4 ranteo rules apply after results are confirmed. However, these are sequential steps (Etapa 3 → Etapa 4), not contradictory preconditions. The flow is internally consistent: ask results in Etapa 3, then rant in Etapa 4 if no results. No shared precondition produces conflicting outputs. |
| BP-5 | CRITICAL | PASS | No cross-file contradictions All SDK files align on behavior and data. Examples.md demonstrates the flow documented in system.md. case_studies.json URLs match those cited in examples.md (e.g., Vivi's URL in Example 1 matches case_studies.json). keywords.json literal_response values align with system.md's resource delivery rules. No file declares behavior or data that contradicts another under the same precondition. |
| BP-6 | MEDIUM | PASS | Voice consistency across examples All 14 examples use consistent rioplatense Argentine Spanish, informal tone, short messages, no emojis, and the same conversational style. Transitions ('claro', 'ok', 'entiendo') and ranteo phrasing are uniform. A reader would recognize the same agent across all examples. |
| BP-7 | CRITICAL | PASS | Unguarded actions (PAT-004) All risky actions carry guards. Etapa 5 (case study) is gated by prior stages (Etapa 3 results confirmation + Etapa 4 ranteo closure with 'se entiende?'). Etapa 6 (videollamada invitation) is gated by lead engagement. Etapa 7 (Calendly link) is gated by explicit lead confirmation ('no, dale', 'sí', 'está bien'). Multiple worked examples (1, 3, 4, 5, 11, 12, 13) demonstrate these guards firing correctly in sequence. |
| BP-8 | HIGH | PASS | No stale version language in examples All example CTAs and closing language match the current flow definition in system.md. Etapa 6 invitation uses the exact phrasing from system.md ('lo mejor que te puedo proponer es una consulta de diagnóstico gratuita...'). Etapa 7 Calendly confirmation uses the exact text ('dale, te paso el link para reservar tu consulta...'). No stale or outdated phrasing detected. |
| BP-9 | MEDIUM | PASS | No duplicated instructions Each behavioral instruction has a single canonical home. The price rule appears once in Principios #1. The 'no internal reasoning' rule appears once in Voz. The Etapa 5 case-study rule appears once in Flujo. No instruction is restated in two locations; cross-references are minimal and appropriate. |
| BP-10 | MEDIUM | PASS | Rule-vs-principle balance Behavioral guidance (tone, transitions, ranteo style) is expressed via Voz section + worked examples. Hard rules are reserved for factual constraints (price in call, Calendly URL, IMC threshold, operating hours). The balance is appropriate: principles + examples govern behavior; rules govern facts and safety. |
| Check | Sec | Estado | Notas |
|---|---|---|---|
| 6a | 1.8 | PASS | NO_RESPONSE present (limits not checked, --no-db) |
| 6c | 3.x | FAIL | 576 lines — rewrite signal |
| 6d | 1.2 | FAIL | 0 examples (target 10, min 7) |
| 6e | 1.3 | N/A | no Principles section |
| 6g | PAT-006 | WARN | non-STEP-0 keyword handling |
| 6h | Anti-pat | WARN | 1 'burbuja' mentions |
| 6i | Anti-pat | PASS | no HTML comments |
| 6j | Anti-pat | FAIL | 30 em-dashes across 1 file(s); first: system.md:15 |
| 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 7 keyword(s) consistent across keywords.json + system.md (no DB) |
| 6s | Hygiene | PASS | no expired dates in live SDK |
| 6t | Coverage | N/A | no Principles or Flujo section found in system.md |
| Fecha | Tema | Tags |
|---|---|---|
| 2026-06-01 | deficit keyword story trigger implementation | feature-request, workflow-automation, instagram-stories, keywords |
| 2026-06-01 | agent activation date and lead volume metrics | metrics, performance, launch |
| 2026-06-01 | weekly funnel report may 25 31 perimenopausia keyword winner | analytics, funnel, keyword-performance, content-strategy |
| 2026-06-01 | instagram profile lookup ingrid ibarra | lead-management, calendly, instagram |
| 2026-06-01 | crm automation expansion lead source tracking and closer con | crm-automation, zapier, lead-qualification, closer-context |
| 2026-05-31 | tipo4 prerequisite bug and workflow fixes | bug, tipo4-follow-ups, workflow-config, prompt-fix |
| 2026-05-31 | tipo1 pre propuesta execution window adjustment | tipo1, workflow-config, execution-window |
| 2026-05-30 | follow up workflow separation tipo3 vs tipo4 | bug-fix, workflow-logic, follow-ups, deployed |
| 2026-05-30 | instagram user lookup calendly reservees | data-lookup, instagram, calendly-sync, manual-work |
| 2026-05-30 | crm automation calendly to sheets | crm, automation, calendly, zapier, proposal |
| 2026-05-29 | workflow follow up multiple messages per slot | bug, workflow, follow-up, specification |
| 2026-05-29 | agent pause resume manual control request | feature-request, agent-control, escalation |
| 2026-05-28 | calendly link message timezone clarification | copy, calendly, preventive, ux, etapa-7 |
| 2026-05-28 | post calendly silence bug logistics messages | bug, calendly, post-agenda, logistics, is_agent_active |
| 2026-05-28 | lead pipeline followup status request | workflow, manual-initiation, agent-activation |
| 2026-05-28 | agent manual conversation activation | workflow, activation, manual-threads, agent-handoff |
| 2026-05-28 | crm automation calendly zapier setup | crm, automation, calendly, zapier, infrastructure |
| 2026-05-28 | prompt fixes and agent activation | deploy, hard-rules, post-calendly-bug, agent-activation, crm-automation, yo-keyword-fix |
| 2026-05-28 | calendly link message timezone clarity | wording, calendly, timezone, friction-reduction, prompt-update |
| 2026-05-28 | post calendly silence logistics messages | bug, post-agenda, logistics, agent-inactive, platform-issue |
| ID | Estado | Cambio pedido |
|---|---|---|
| B9 | pendiente | Reemplazar referencias a "burbuja" en system.md por "mensaje aparte" — el LLM no entiende el concepto plataforma de burbujas |
| C1 | pendiente | **Etapa 1 — Abridor**: variar según fuente (hand raiser recurso entregado/pendiente, story, CTA INFO/CAMBIO/YO, recurso por keyword, bienvenida nuevo follower, DM directo <30d, DM |
| X8 | pendiente | Nurturing score 46.9 (sintéticos) — bajo, pendiente mejora |
| X9 | pendiente | Flow score 77.5 (sintéticos) — pendiente mejora con data real |
| Señal | Valor | Evidencia (cita del cliente) |
|---|---|---|
| Enojado con la plataforma | no | — |
| Es queja | sí | — |
| Queja constructiva (accionable) | sí | — |
| Señal | Cita textual | Fecha |
|---|---|---|
| El cliente está activamente operando y optimizando su funnel, señal de engagement positivo | “None” | 2026-06-01 |
| El cliente solicita feature de enriquecimiento de CRM con contexto previo a llamada, indicando que ve valor en la plataforma y quiere escalar | “None” | 2026-06-01 |
| El cliente solicita nuevo keyword trigger para story replies (deficit), mostrando iniciativa de expansión | “None” | 2026-06-01 |
| El cliente reportó bug de mensajes spam (25 mensajes) — queja accionable sin descarga emocional visible en los digests | “None” | 2026-05-30 |
| El cliente solicitó ajuste de ventana de ejecución y condición de guardia en tipo1, señal de que monitorea de cerca el comportamiento del agente | “None” | 2026-05-31 |
| El cliente consulta métricas de activación y volumen de leads, comportamiento típico de usuario enganchado con resultados | “None” | 2026-06-01 |
Un creador que pide cambios concretos está enganchado (mejor señal que el silencio).
| ID | Estado | Cambio pedido |
|---|---|---|
| B9 | pendiente | Reemplazar referencias a "burbuja" en system.md por "mensaje aparte" — el LLM no entiende el concepto plataforma de burbujas |
| C1 | pendiente | **Etapa 1 — Abridor**: variar según fuente (hand raiser recurso entregado/pendiente, story, CTA INFO/CAMBIO/YO, recurso por keyword, bienvenida nuevo follower, DM directo <30d, DM |
| X8 | pendiente | Nurturing score 46.9 (sintéticos) — bajo, pendiente mejora |
| X9 | pendiente | Flow score 77.5 (sintéticos) — pendiente mejora con data real |