feat(cmo): add in-house X automation baseline, queue review gate, and docs

Author: ccross2

ops/cmo-automation/.gitignore

@@ -0,0 +1,10 @@
+# Generated runtime artifacts
+data/*.json
+reports/cmo-analysis.json
+reports/cmo-analysis.md
+reports/cmo-queue-review.json
+reports/cmo-queue-review.md
+
+# Python cache
+__pycache__/
+*.pyc

ops/cmo-automation/README.md

@@ -0,0 +1,53 @@
+CMO Automation (X/Twitter)
+
+Purpose
+- Internal, data-driven replacement for outsourced engagement operations.
+- Covers three coordinated accounts:
+  - @TheCesarCross
+  - @sovren_software
+  - @mrhaven_agent
+
+What this does today
+1) Collect snapshot data from X API (profile + timeline)
+2) Analyze behavior and output baseline metrics
+3) Generate a conservative dry-run engagement queue
+4) Score the queue for risk/quality before any execution
+
+Current mode
+- Dry-run / assisted only.
+- No autonomous posting in this module yet.
+
+Structure
+- config/cmo_accounts.yaml
+- scripts/collect_x_data.py
+- scripts/analyze_x_cmo.py
+- scripts/generate_engagement_queue.py
+- scripts/review_engagement_queue.py
+- reports/CMO-AUTOMATION-IMPLEMENTATION-PLAN.md
+
+Quick start
+1) Export credentials (or source ~/.claude/secrets.env)
+   - Required: X_API_KEY, X_API_SECRET, X_BEARER_TOKEN, X_ACCESS_TOKEN, X_ACCESS_TOKEN_SECRET
+   - Script also maps from TWITTER_* variables.
+2) Run:
+   - python3 scripts/collect_x_data.py
+   - python3 scripts/analyze_x_cmo.py
+   - python3 scripts/generate_engagement_queue.py
+   - python3 scripts/review_engagement_queue.py
+
+Outputs
+- data/latest.json
+- reports/cmo-analysis.json
+- reports/cmo-analysis.md
+- data/engagement-queue.json
+- reports/cmo-queue-review.json
+- reports/cmo-queue-review.md
+
+Guardrails
+- Generic short-reply behavior is penalized.
+- Repetitive opener patterns are penalized.
+- Per-account recommended caps are enforced by policy review before execution.
+
+Notes
+- Generated JSON/analysis artifacts are ignored in git by default.
+- Commit strategy/docs/scripts, not volatile runtime snapshots.

ops/cmo-automation/config/cmo_accounts.yaml

@@ -0,0 +1,42 @@
+accounts:
+  - handle: TheCesarCross
+    role: founder
+    voice: founder-authentic
+    target_mix:
+      root_posts_per_day: 1
+      replies_per_day: 12
+      quote_posts_per_week: 4
+  - handle: sovren_software
+    role: brand
+    voice: thesis-terse
+    target_mix:
+      root_posts_per_day: 1
+      replies_per_day: 10
+      quote_posts_per_week: 3
+  - handle: mrhaven_agent
+    role: product-agent
+    voice: guardian-utility
+    target_mix:
+      root_posts_per_day: 2
+      replies_per_day: 6
+      quote_posts_per_week: 2
+
+analysis:
+  lookback_days: 21
+  baseline_label: fiverr-period
+  detect_patterns:
+    short_reply_threshold_chars: 90
+    short_reply_ratio_alert: 0.75
+    repeated_openers:
+      - "Totally agree"
+      - "Agree"
+      - "Appreciate"
+      - "Nice take"
+      - "We value"
+      - "Right then"
+
+risk_controls:
+  max_replies_per_hour_per_account: 8
+  random_delay_seconds: [45, 180]
+  avoid_repeated_same-user-replies_within_hours: 24
+  dry_run_default: true

ops/cmo-automation/reports/CMO-AUTOMATION-IMPLEMENTATION-PLAN.md

@@ -0,0 +1,159 @@
+# Automated CMO Function — Implementation Plan (Three X Accounts)
+
+Goal
+- Replace outsourced Fiverr engagement with an in-house, measurable, low-risk automation system that coordinates:
+  - @TheCesarCross (founder)
+  - @sovren_software (brand)
+  - @mrhaven_agent (product agent)
+
+Current baseline (from latest captured sample)
+- @TheCesarCross: 7/7 posts are replies, short-reply ratio 0.714, avg impressions 3.43
+- @sovren_software: 8/8 posts are replies, short-reply ratio 0.875, avg impressions 6.25
+- @mrhaven_agent: 8/8 posts are root posts, avg impressions 5.62
+- Pattern observed in outsourced-style behavior: high-volume short replies, broad low-overlap targeting, weak engagement yield.
+
+Interpretation
+- The current engagement mode is over-indexed on low-context reactive replies.
+- Root narrative production is underrepresented on founder + brand accounts in sampled window.
+- This aligns with low return on outsourced spend.
+
+---
+
+## Phase 1 — Instrumentation + Baseline Lock (now)
+
+Implemented
+- Collector: scripts/collect_x_data.py
+- Analyzer: scripts/analyze_x_cmo.py
+- Action queue generator: scripts/generate_engagement_queue.py
+- Config: config/cmo_accounts.yaml
+
+Output artifacts
+- data/latest.json
+- reports/cmo-analysis.json
+- reports/cmo-analysis.md
+- data/engagement-queue.json
+
+Operating command
+- cd ~/cDesign/sovren-website/ops/cmo-automation
+- source ~/.claude/secrets.env
+- export CMO_TIMELINE_MAX=20
+- python3 scripts/collect_x_data.py
+- python3 scripts/analyze_x_cmo.py
+- python3 scripts/generate_engagement_queue.py
+
+---
+
+## Phase 2 — Reconstruct Fiverr Playbook (1–2 days)
+
+Method
+1) Pull daily snapshots for 14 days.
+2) Tag candidate outsourced actions with heuristics:
+   - starts with @
+   - <= 90 chars
+   - generic opener patterns
+   - low dwell (burst cadence)
+3) Split output into:
+   - worked: replies with above-median impressions/likes
+   - failed: replies with 0 interaction and low impressions
+
+Deliverables
+- reports/fiverr-playbook-reconstruction.md
+- reports/fiverr-pattern-confusion-matrix.json
+
+Decision gate
+- If short reply ratio > 0.65 and avg likes < 0.5 on targeted accounts, cap automated replies and shift volume to root + quote strategy.
+
+---
+
+## Phase 3 — Improved Strategy Model (systematized)
+
+Account strategy
+1) Founder (@TheCesarCross)
+- Primary: authority and framing
+- Mix target: 35% root, 50% high-signal replies, 15% quotes
+- Rule: no generic acknowledgements without differentiated opinion
+
+2) Brand (@sovren_software)
+- Primary: thesis + product context
+- Mix target: 45% root, 40% contextual replies, 15% quotes
+- Rule: every reply should ladder to sovereignty thesis or product proof
+
+3) Product (@mrhaven_agent)
+- Primary: utility + proofs
+- Mix target: 60% root utility posts, 25% proof/context replies, 15% quote amplification
+- Rule: keep proof-linked clarity; avoid broad-topic drift
+
+Cross-account orchestration rules
+- Founder seeds perspective → brand codifies thesis → mrhaven_agent supplies proof/utility.
+- No duplicate same-angle reply across two accounts within 12 hours.
+- Daily coordination cap: 1 shared topic cluster/day.
+
+---
+
+## Phase 4 — Automation Execution Layer (safe automation)
+
+Execution architecture
+- Scheduler: cron (or GitHub Actions)
+- Data: snapshot JSON + analysis report + queue JSON
+- Executor modes:
+  - dry-run (default): generate recommended actions only
+  - assisted: human approves queued actions
+  - auto-lite: post only pre-approved templates with strict limits
+
+Risk controls
+- per-account reply caps/hour
+- 24h cooldown on same-user repeated replies
+- random delay jitter
+- require contextual relevance score before posting
+- hard block list + topic deny list
+
+Immediate policy change
+- pause blind high-volume replying
+- enforce minimum root-post floor each account/day
+- route all auto actions through queue + review for first 2 weeks
+
+---
+
+## Phase 5 — KPI and Feedback Loop
+
+North-star KPIs
+- engagement per post by type (root/reply/quote)
+- follower delta/week
+- reply conversion rate (reply -> profile visit/follow)
+- cross-account amplification uplift
+
+Guardrail KPIs
+- % low-signal replies (short generic) < 30%
+- duplicate semantic replies/day < 10%
+- actions rejected by reviewer (quality failure)
+
+Weekly review output
+- reports/weekly-cmo-review-YYYY-MM-DD.md
+- includes: what changed, what worked, what to stop, next week experiments
+
+---
+
+## What to do next (execution order)
+
+1) Start daily collection
+- Run collector+analyzer twice daily for 14 days.
+
+2) Add first automation gate
+- Keep generate_engagement_queue.py in dry-run.
+- Review queue manually and execute only approved actions.
+
+3) Build reply-quality scorer
+- Add lightweight semantic rubric:
+  - specificity
+  - thesis alignment
+  - non-generic value add
+
+4) Promote to auto-lite only after 2-week KPI check
+- If KPIs improve and guardrails hold, allow bounded auto execution.
+
+---
+
+Status
+- Baseline system scaffold is implemented under:
+  ~/cDesign/sovren-website/ops/cmo-automation
+- Ready for day-1 operation and playbook reconstruction.