mancitrus/system-prompts-and-models-of-ai-tools
1
0
Fork 0
mirror of https://github.com/x1xhlol/system-prompts-and-models-of-ai-tools.git synced 2026-06-17 23:09:35 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
601 Commits 1 Branch 0 Tags 81 MiB
e106a9a0d2
Commit Graph

6 Commits

Author SHA1 Message Date
Dealix Builder
e106a9a0d2 feat(targeting+service+excellence): Saudi Targeting OS + Service Tower + Service Excellence — 38 modules + 62 endpoints + 105 tests 
Targeting & Acquisition OS (16 modules) — نظام الاستهداف الذكي
- account_finder: account-first targeting; 12 buying signals; deterministic 10-25 accounts per (sector, city)
- buyer_role_mapper: 14 buyer roles + sector-specific buying-committee maps + role-based Arabic angles
- contact_source_policy: 12 sources (crm_customer→opt_out) with risk_score, channels-allowed, retention guidance, lawful_basis
- contactability_matrix: 5 action modes (suggest_only/draft_only/approval_required/approved_execute/blocked); opt-out always blocked
- linkedin_strategy: Lead Forms + Ads + manual ONLY; linkedin_do_not_do() locks scrape/auto-DM/auto-connect/extensions
- email_strategy: drafts + unsubscribe footer + domain-pacing (fresh/warmed/trusted/damaged) + spam-trigger risk
- whatsapp_strategy: opt-in only; rejects cold + risky phrases; opt-in template requires explicit purpose+company+unsubscribe
- social_strategy: official APIs only; listening + drafts; no auto-publish
- outreach_scheduler: day-by-day plans + daily limits + opt-out enforcement
- reputation_guard: bounce/complaint/opt-out thresholds → healthy/watch/pause + recovery actions per channel
- daily_autopilot: Arabic brief + 7 today actions + EOD report
- acquisition_scorecard: pipeline + meetings + risks + productivity_score
- self_growth_mode: 5 ICP focuses for Dealix; daily brief + monthly targets
- free_diagnostic: Free Growth Diagnostic (3 ops + msg + risk + plan) → paid pilot recommendation
- contract_drafts: Pilot/DPA/Referral/Agency/SOW outlines (legal_review_required, PDPL-aware)
- service_offers: 7 targeting-tier offers + recommend by customer-type

Service Tower (8 modules) — برج الخدمات الذاتية (12 productized services)
- service_catalog: 12 services with target_customer/outcome/inputs/workflow/deliverables/pricing/risk/proof/upgrade
- service_wizard: deterministic recommend (agency→partner; list→list_intelligence; founder→self_growth; CEO→exec_brief; budget≥2999→growth_os; default→first_10)
- mission_templates: workflow steps with approval gates + linked growth missions
- pricing_engine: SAR quotes scaled by company_size×urgency×channels_count + setup_fee + monthly_offer
- deliverables: client report outline + proof pack template + operator checklist (no live actions)
- service_scorecard: 0..100 score from drafts/replies/meetings/pipeline/CSAT
- whatsapp_ceo_control: daily brief, approval cards (≤3 buttons), risk alerts, EOD reports
- upgrade_paths: deterministic next-service recommendation + Arabic upsell messages

Service Excellence OS (8 modules) — مصنع الخدمات الممتازة
- feature_matrix: 12 must-have features per service + advanced/premium/future tiers
- service_scoring: 10-dimension excellence score (clarity, speed_to_value, automation, compliance, proof, upsell, uniqueness, scalability, ops_daily, proof_data) → launch_ready/beta_only/needs_work
- quality_review: 4 gates (proof / approval / pricing / channels) + status verdict; review_service_before_launch and review/all
- competitor_gap: 7 competitor categories (CRM, WhatsApp tools, email assistants, LinkedIn tools, agencies, revenue intelligence, generic AI) + Dealix advantages + do-not-copy
- proof_metrics: required metrics + ROI estimate (pipeline_x + closed_won_x) + Arabic summary
- research_lab: monthly brief + feature hypotheses + top-3 experiments + monthly review
- service_improvement_backlog: feedback→backlog conversion + impact/effort prioritization + weekly improvements
- launch_package: landing outline + sales script + 12-min demo script + 5-day onboarding checklist

Routers (3 new) — 62 endpoints
- /api/v1/targeting/* — 20 endpoints (accounts, buying-committee, contacts, uploaded-list, outreach, daily-autopilot, self-growth, reputation, linkedin, drafts, free-diagnostic, services, contracts)
- /api/v1/services/* — 20 endpoints (catalog, recommend, intake, start, workflow, deliverables, proof-pack, quote, setup-fee, monthly-offer, scorecard, upgrade-path, ceo daily-brief/approval-card/risk-alert/EOD)
- /api/v1/service-excellence/* — 22 endpoints (feature-matrix, score, quality-review, review/all, proof-metrics, roi-estimate, gap-analysis, research-brief, hypotheses, experiments, monthly-review, backlog, weekly-improvements, launch-package, landing/sales/demo/onboarding)

Tests (3 new files, 105 tests)
- test_targeting_os: 47 tests (Arabic accounts, buying committees, opt-out blocked, cold WA blocked, LinkedIn no-scraping, email unsubscribe, WA risk, outreach plan, reputation guard, self-growth, contracts, scorecard)
- test_service_tower: 38 tests (12+ services, all have pricing/proof/deliverables/approval, wizard recommendations, workflow includes approval, quote scales, CEO cards ≤3 buttons, no live send)
- test_service_excellence: 33 tests (feature matrix, score returns status, ALL services pass quality gates, ROI x-multiples, 7 competitor categories, hypotheses+experiments, backlog conversion, launch package complete, demo=12min)

Docs (3 new + 1 updated)
- TARGETING_ACQUISITION_OS.md (Arabic)
- SERVICE_TOWER_STRATEGY.md (Arabic)
- SERVICE_EXCELLENCE_OS.md (Arabic)
- DEALIX_100_PERCENT_LAUNCH_PLAN.md — added §36 Targeting OS + §37 Service Tower + §38 Service Excellence + §39 Landing Pages

Landing pages (4 new, RTL Arabic)
- services.html — 3 doors + 12 productized services
- free-diagnostic.html — free growth diagnostic
- first-10-opportunities.html — kill feature
- agency-partner.html — agency partner program

Test results
- 105/105 new tests pass
- Full suite: 768 passed, 2 skipped
- 0 existing tests broken

Safety + integration with previous layers
- Targeting OS uses contactability_matrix → ALL contacts gated before any send
- Service Tower's workflow includes approval gate; ALL services live_send_allowed=False
- Service Excellence quality_review BLOCKS launch on missing proof/approval/pricing/unsafe channels
- linkedin_do_not_do() encodes 8 explicit prohibitions (scraping/auto-DM/auto-connect/extensions)
- whatsapp_do_not_do() blocks cold sends + group scraping
- Contracts always: legal_review_required=True, not_legal_advice=True, PDPL sections present
- Self-Growth Mode lets Dealix target its OWN ICP using the same approval-first pipeline

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-01 17:11:00 +03:00
Dealix Builder
bcf545c22e feat(self-improving): Hermes-inspired Agent Platform — 6 layers + 30 endpoints + 76 tests + Private Beta launch 
Security Curator (4 modules) — جدار الحماية الأول
- secret_redactor: 11 patterns (GitHub PAT, OpenAI/Anthropic/Supabase/WhatsApp/Moyasar/Sentry/Google/AWS/private keys); never returns raw secret
- patch_firewall: blocks .env / credentials.json / RSA keys; scans added lines for secret patterns
- trace_redactor: masks phones (+966...) and emails for PII safety
- tool_output_sanitizer: cleans tool outputs before they hit ledger/Proof Pack/UI/observability

Growth Curator (5 modules) — التحسين الذاتي
- message_curator: grades Arabic messages (0..100), detects 8 risky phrases, suggests Saudi-tone skeleton
- playbook_curator: scores playbooks by outcome (accept/reply/meeting/deal); winner/promising/needs_work/archive
- mission_curator: scores completed missions; ship_it_widely/iterate/rework_or_retire
- skill_inventory: deterministic 23-skill catalog across 5 layers
- curator_report: weekly Arabic summary "ماذا تعلمنا هذا الأسبوع"

Meeting Intelligence (5 modules) — ذكاء الاجتماعات
- transcript_parser: accepts Google Meet entries OR plain "Speaker: text" format
- meeting_brief: 6-section pre-meeting brief in Arabic (objective/questions/objections/offer/next-step)
- objection_extractor: 8 categories (price/timing/authority/trust/integration/competitor/results/complexity)
- followup_builder: email + WhatsApp drafts; live_send_allowed=False always
- deal_risk: 0..100 score from objections + missing next-step + decision-maker absence + days-since-touch

Model Router (5 modules) — موجّه النماذج
- provider_registry: 7 providers (Claude Sonnet/Haiku, GPT-4-class, GPT-4o-mini, Gemini Pro, Azure OAI KSA-region, Local Qwen Arabic-tuned)
- task_router: 10 task types × routing decisions with reasons_ar
- cost_policy: bulk → low; output > 1500 tokens → high
- fallback_policy: high-sensitivity workloads prefer KSA-region/self-hosted FIRST
- usage_dashboard: deterministic demo of all task routes

Connector Catalog (3 modules) — كتالوج التكاملات
- 14 connectors (WhatsApp Cloud, Gmail, Calendar, Google Meet, Moyasar, LinkedIn Lead Forms, Google Business Profile, X API, Instagram, Sheets, CRM, Website Forms, Composio, MCP Gateway)
- Each has launch_phase (1-4), risk_level, allowed_actions, blocked_actions, Arabic risk dossier
- WhatsApp blocks cold_send_without_consent; Moyasar blocks store_card_number; MCP requires allowlist

Agent Observability (5 modules) — مراقبة الوكلاء + التقييمات
- trace_events: SHA256-hashes user/company IDs; sanitizes payload/output before logging
- safety_eval: 7 rules (guarantee, scarcity_fake, medical_claim, financial, regulatory, personal_data, urgency); 0..100 → safe/needs_review/blocked
- saudi_tone_eval: positive markers (هلا, لاحظت, يناسبك) vs negative (تحية طيبة وبعد, synergy, leverage); arabic_ratio bonus
- eval_pack: 5 curated cases with expected verdicts
- cost_tracker: per workflow/provider/task_type aggregation

Routers (6 new) — 30 endpoints
- /api/v1/security-curator/{demo, redact, inspect-diff, sanitize-output}
- /api/v1/growth-curator/{skills/inventory, messages/grade, messages/improve, messages/duplicates, missions/next, report/weekly, report/demo}
- /api/v1/meeting-intelligence/{brief, brief/demo, transcript/summarize, followup/draft, deal-risk}
- /api/v1/model-router/{providers, tasks, route, cost-class, usage/demo}
- /api/v1/connector-catalog/{catalog, summary, status, risks, {key}}
- /api/v1/agent-observability/{trace/build, safety/eval, tone/eval, evals/run}

Tests (6 new files, 76 tests)
- test_security_curator: 16 tests (PAT detect, key redact, env diff block, payload scan, trace mask)
- test_growth_curator: 16 tests (Arabic grade, risky phrases, dup detect, playbook scoring, mission recommend, weekly report)
- test_meeting_intelligence: 13 tests (transcript parse, brief sections, objection extract, followup drafts, deal risk)
- test_dealix_model_router: 11 tests (every task → ≥1 provider, KSA-region for high sensitivity, cost class, primary override)
- test_agent_observability: 12 tests (trace hashing, safety verdicts, tone scoring, eval pack)
- test_connector_catalog: 11 tests (≥12 connectors, every has risk/blocked actions, WA cold-send blocked, Moyasar card-storage blocked)

Docs (8 new + 1 updated)
- AGENT_SECURITY_CURATOR.md (Arabic)
- GROWTH_CURATOR_STRATEGY.md (Arabic)
- MEETING_INTELLIGENCE.md (Arabic)
- MODEL_PROVIDER_ROUTER.md (Arabic)
- CONNECTOR_CATALOG.md (Arabic)
- AGENT_OBSERVABILITY_EVALS.md (Arabic)
- PRIVATE_BETA_LAUNCH_TODAY.md (Arabic) — go-checklist + offer + risks
- DEMO_SCRIPT_12_MINUTES.md (Arabic) — minute-by-minute demo flow
- FIRST_20_OUTREACH_MESSAGES.md (Arabic) — 7 personas + 3 follow-ups, all under safety/tone evals
- DEALIX_100_PERCENT_LAUNCH_PLAN.md — added §34 Self-Improving Agent Platform + §35 Private Beta Launch

Landing
- landing/private-beta.html — Arabic RTL, dark theme, pricing, 11 demo endpoints, safety banner

Test results
- 76/76 new tests pass
- Full suite: 663 passed, 2 skipped (missing API keys, unrelated)
- 0 existing tests broken

Safety
- All 6 layers honor approval-first, draft-only, no-live-send
- Hash user/company IDs before any trace
- No secrets in logs/embeddings/traces (3-layer defense: redactor + sanitizer + firewall)
- Saudi tone eval rejects "تحية طيبة وبعد" + "synergy" auto-corporate language
- Safety eval blocks "ضمان 100%" + medical claims + fake urgency
- Connector Catalog: WhatsApp blocks cold-send, Moyasar blocks card storage, MCP requires allowlist

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-01 16:30:18 +03:00
Dealix Builder
4e969131c7 feat(platform+intelligence): Growth Control Tower + Growth Neural Network — 20 modules + 25 endpoints + 60 tests 
Platform Services Layer (10 modules) — برج التحكم بالنمو
- event_bus: 27 typed events (whatsapp/email/calendar/lead/payment/review/social/partner/sheet/crm/action)
- identity_resolution: cross-channel merge (phone+email+CRM+social) with confidence scoring
- channel_registry: 11 channels (WA, Gmail, Calendar, Moyasar, LinkedIn, X, IG, GBP, Sheets, CRM, Forms) with capabilities/risk/PDPL notes
- action_policy: 9 rules (block_cold_whatsapp, block_payment_no_confirm, block_secrets, external_send_needs_approval, calendar_insert_needs_approval, social_dm_needs_explicit, unknown_source_review, high_value_deal_review, draft_only_safe)
- tool_gateway: single execution chokepoint, env-flag-gated live actions (default OFF)
- unified_inbox: 8 card types, ≤3 buttons enforced, Arabic
- action_ledger: requested→approved→executed audit trail
- proof_ledger: leads/meetings/drafts/sends/payments/revenue/risks_blocked/time_saved per channel
- service_catalog: 12 sellable services
- router api/routers/platform_services.py — 13 endpoints under /api/v1/platform/

Intelligence Layer (10 modules) — الشبكة العصبية للنمو
- growth_brain: per-customer Brain + is_ready_for_autopilot() (≥30 signals + ≥40% accept)
- command_feed: 9 daily card types (opportunity/revenue_leak/partner_suggestion/meeting_prep/review_response/competitive_move/customer_reactivation/ai_visibility_alert/action_required)
- action_graph: 10 typed edges (signal→action→outcome) with what_works_summary
- mission_engine: 7 missions, KILL FEATURE first_10_opportunities (10 فرص في 10 دقائق)
- decision_memory: learns from accept/skip/edit/block, returns preferences (channels, tones, sectors, rejected actions, accept_rate)
- trust_score: composite 0-100 (source+opt_in+channel+content+freq+approval) → safe/needs_review/blocked
- revenue_dna: best_channel/segment/angle + common_objection + avg_cycle_days
- opportunity_simulator: 9 Saudi sectors, expected_replies/meetings/deals/pipeline_sar + risk_score
- competitive_moves: 8 move types with Arabic recommended_action_ar
- board_brief: weekly Founder Shadow Board (3 decisions + 3 opportunities + 3 risks + relationship + experiment + metric)
- router api/routers/intelligence_layer.py — 12 endpoints under /api/v1/intelligence/

Tests
- tests/unit/test_platform_services.py — 31 tests covering catalog/channels/events/policy/gateway/identity/inbox/ledger/proof
- tests/unit/test_intelligence_layer.py — 29 tests covering brain/feed/graph/missions/memory/trust/dna/simulator/competitive/brief
- 60/60 new tests pass; full suite 587 passed, 2 skipped

Docs
- docs/PLATFORM_SERVICES_STRATEGY.md (Arabic)
- docs/INTELLIGENCE_LAYER_STRATEGY.md (Arabic)
- docs/DEALIX_100_PERCENT_LAUNCH_PLAN.md — added §32 Platform Services + §33 Intelligence Layer

Safety
- No live send by default (all WA/Gmail/Calendar/Moyasar guarded by env flags, all OFF)
- All external actions go through Tool Gateway → Action Policy → draft/approval_required
- No secrets allowed in payloads (block_secrets policy)
- PDPL-aware: cold WhatsApp without consent is hard-blocked
- Existing 477+ tests untouched (no breaking changes)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-01 16:05:12 +03:00
Dealix Builder
8942c6e84c feat(growth-operator): Arabic Growth Operator — 10 modules + 16 endpoints + 50 tests 
Builds the full Saudi Autonomous Revenue OS surface as 10 deterministic
modules + a 16-endpoint router under /api/v1/growth-operator/.
Approval-first: every outbound is draft. No live send / charge / calendar
insert from this layer.

MODULES (auto_client_acquisition/growth_operator/)

1. client_profile.py — ClientGrowthProfile + Saudi-default approval
   + compliance rules (no cold WhatsApp, blocked keywords, weekly cap,
   quiet_hours_riyadh)
2. contact_importer.py — normalize_phone (Saudi E.164),
   dedupe_contacts (richer-record-wins), classify_contact_source
   (existing/inbound/event/referral/old_lead/cold/unknown), detect_opt_out
   (Arabic + English markers), summarize_import (dashboard report)
3. contactability.py — score_contactability returns
   safe/needs_review/blocked with Arabic reasons; default policy:
   no cold WhatsApp without lawful basis (PDPL Art.5)
4. targeting.py — segment_contacts, rank_targets (filters unsafe),
   recommend_top_10, why_now_stub (deterministic, sector-aware)
5. message_planner.py — draft_arabic_message (Saudi tone, 4-sector
   opener bank, no overhyped phrases, always pending_approval),
   draft_followup (4 outcome modes), draft_objection_response
   (6 indexed Saudi B2B objections with score_delta + next_action)
6. partnership_planner.py — 6 partner types catalog
   (agency / consultant / integrator / crm / community / influencer)
   + suggest_partner_types (size/sector aware) + draft_partner_outreach
   + partner_scorecard (platinum/gold/silver/bronze)
7. meeting_planner.py — build_meeting_agenda (15/20-30/45+ min slot
   plans), build_calendar_draft (Google Calendar shape, live_inserted=False,
   conferenceData for Meet, Asia/Riyadh timezone), build_post_meeting_followup
8. payment_offer.py — sar_to_halalas, build_moyasar_payment_link_draft
   (full payload + in-chat message + 4-plan catalog, live_charged=False)
9. proof_pack.py — build_weekly_proof_pack with grade A+/A/B/C/D,
   activity/money/quality/best-of sections, dynamic next_week_plan_ar,
   markdown export
10. mission_planner.py — 6 GROWTH_MISSIONS (first_10_opportunities ⭐
    kill feature, recover_stalled_deals, partnership_sprint,
    safe_whatsapp_campaign, meeting_booking_sprint, list_cleanup);
    list_missions() + run_mission()

ROUTER (api/routers/growth_operator.py) — 16 endpoints

POST /contacts/import-preview · POST /contactability/score
POST /targets/top-10 · POST /messages/draft · POST /messages/followup
POST /messages/objection-response · POST /partners/suggest
POST /partners/outreach · POST /partners/scorecard
POST /meetings/draft · POST /meetings/post-followup
POST /payment-offer/draft · GET /missions · POST /missions/{id}/run
GET /proof-pack/demo · POST /profile

WIRING: api/main.py adds growth_operator import + router include
(positioned after personal_operator, before public).

DOCS

- docs/ARABIC_GROWTH_OPERATOR_FULL_SPEC.md (NEW): 20-section vision +
  customer-type table + upload flow + contactability rules +
  WhatsApp/Gmail/Calendar/Moyasar drafts + 6 missions + 16-endpoint
  catalog + competitive comparison + beta readiness checklist

TESTS — 50 passing on Python 3.10 venv

tests/unit/test_growth_operator.py covers:
- Phone normalization across 5 input formats including invalid
- Dedupe richer-record invariant
- Source classification (existing/inbound/event/cold/unknown)
- Opt-out detection (Arabic + English notes + status)
- Import summary aggregation
- Contactability: opt-out blocked, cold WhatsApp blocked,
  unknown→needs_review, existing→safe, inbound→safe
- Bulk contactability summary
- Top-10 filtering (unsafe excluded), max-cap enforcement
- Segment buckets
- Arabic message: pending_approval invariant + Arabic content
  + no overhyped phrases (banned list)
- Followup approval invariant
- Objection response: known + unknown→diagnostic
- Partner suggestions size-aware (SMB→agency/consultant/community)
- Partner outreach approval invariant
- Partner unknown type returns error
- Partner scorecard tier ordering
- Meeting agenda + calendar draft (live_inserted=False) +
  Asia/Riyadh timezone + post-followup pending
- Payment: halalas conversion (1 SAR=100), negative raises,
  draft NEVER charges (live_charged=False), unknown plan→error
- Proof pack: grade range + structure + markdown export
- Missions: first_10_opportunities present + kill feature ID
  + run mission known/unknown
- Profile: demo specialized + partial not specialized
  + default compliance blocks 'ضمان 100' + no_cold_whatsapp_without_lawful_basis

VERIFICATION
- 527 unit tests pass (was 477; +50 growth_operator)
- 2 skipped (provider smoke needs API keys)
- AST green on all 13 new files
- Approval invariant holds across every drafting function

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-01 15:33:11 +03:00
Dealix Builder
e34cc729aa feat(dealix): py3.10/3.11 compat shim + 54 unit tests for business/innovation/ai 
PROBLEM
The codebase used Python 3.11+ stdlib features (`from datetime import UTC`,
`from enum import StrEnum`) in 22 files, breaking local dev on Python 3.10
(Windows users) and any pytest run that imports the affected modules.

SOLUTION
1. New `core/_py_compat.py` providing UTC + StrEnum shims that:
   - On 3.11+ re-export the stdlib names (zero overhead)
   - On 3.10 fall back to `timezone.utc` and a (str, Enum) backport

2. All 22 affected files patched to import from the shim:
   - core/utils.py, core/config/models.py
   - api/routers/admin.py
   - auto_client_acquisition/{ai/model_router, agents/{intake,icp_matcher},
     v3/{memory,agents,compliance_os,market_radar},
     personal_operator/{operator,memory,launch_report},
     innovation/{proof_ledger_repo,command_feed_live}}.py
   - autonomous_growth/agents/sector_intel.py
   - dealix/{trust/{approval,tool_verification,policy},
     observability/cost_tracker,
     contracts/{evidence_pack,event_envelope,audit_log,decision},
     classifications/__init__,
     governance/approvals}.py

3. Three new test suites for previously-untested layers (54 tests):
   - tests/unit/test_business_suite.py — gtm_plan, launch_metrics,
     market_positioning, pricing_strategy, proof_pack, unit_economics,
     verticals (28 tests covering plan recommendation, performance fee,
     ROI math, account health grading, vertical playbook structure)
   - tests/unit/test_innovation_suite.py — aeo_radar, command_feed,
     deal_rooms, experiments, growth_missions, proof_ledger, ten_in_ten
     (18 tests covering deterministic reproducibility, card type taxonomy,
     pending-approval invariant, kill-mission visibility)
   - tests/unit/test_ai_model_router.py — ModelTask + get_model_route +
     estimate_model_cost_class + requires_guardrail (8 tests covering
     enum integrity, route round-trip, guardrail bool contract)

VERIFICATION
- ast.parse green on all 22 patched files
- pytest tests/unit/ → 477 passed, 2 skipped (provider smoke needs API keys)
  on Python 3.10.12 venv with project requirements installed
- No behavior change on 3.11+: the shim re-exports stdlib symbols

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
 2026-05-01 14:50:04 +03:00
Sami Assiri
f79c69ff25 ci(dealix): root GitHub workflows, ai-company track, full Dealix API tree 
Made-with: Cursor
 2026-05-01 14:03:52 +03:00