clawd/memory/2026-02-14.md

11 KiB
Raw Blame History

2026-02-14 (Saturday) — Valentine's Day

Context Hygiene Rules Added to AGENTS.md

  • Side questions → subagent always (keep main context clean)
  • Both units always (metric + imperial) — Johan's brain is metric, lives in US
  • Thinking level: match the task (no thinking for simple stuff)
  • Also added units preference to USER.md

myCigna — Autonomous Login Achievement

  • Installed real Google Chrome on forge (v145), runs on Xvfb:99 with CDP on port 9224
  • Headless Playwright gets WAF-blocked by Cigna; real Chrome on Xvfb works perfectly
  • Login flow: username=tjjongsma (NOT email), password in TOOLS.md
  • 2FA via email to tj@jongsma.me — grabbed code from MC automatically, zero human intervention
  • Full autonomous flow: login → 2FA → navigate → extract claims data
  • Credentials stored in TOOLS.md under "myCigna"

Baycare Ventilator Fraud (CRITICAL — Sophia)

  • Baycare Home Care billing HCPCS E0465 (home ventilator) at $3,125/month
  • Sophia does NOT have a ventilator. Never had one from Baycare. Off vent since Nov 2022.
  • Jan 10 claim #260155327900: $3,125 billed, $1,781.25 applied to deductible
  • Feb 10 claim #260540751600: $3,125 billed, still processing (also E0465)
  • Other Baycare charge E0431 ($121.60/mo) = portable oxygen rental — legitimate
  • Johan has notified both Baycare and Cigna at least 3 times before
  • Family is extremely self-supporting: concierge pediatrician, own suppliers (Tri-Med, Amazon), own portable O2 (Zen-O)
  • Baycare only provides: consumable supplies (gauze, trach ties, suction tubes, gloves, H2O2, etc.)
  • The ONLY equipment Baycare could legitimately bill for = suction machine (~$300)
  • Strategy: Don't pay. Let them escalate. Documentation ready.
  • Formal complaint drafted: ~/documents/records/medical/baycare-ventilator-fraud-complaint-2026-02-14.md
  • If ~38 months of E0465 since Nov 2022: potentially ~$118,750 in fraudulent charges

Cron Job Finalization

  • Git audit timeout bumped 120s → 300s
  • Evening Work Queue → Evening Briefing (now isolated agentTurn, same format as morning)
  • Both briefings now require: actual release notes/changelog with links for any software updates
  • Both briefings now include infrastructure/industry topics: memory pricing, HDD pricing, hyperscaler price changes, backup/MSP industry (Veeam, Commvault, Cohesity, Rubrik, NinjaOne, AvePoint, NABL, etc.)
  • Final cron count: 7 jobs (claude-usage, git-audit, morning briefing, evening briefing, nightly maintenance, weekly synthesis, Dr. Madan reminder)

Shannon Security Scanner — Live on Zurich VPS

  • Configured Claude Code Router (CCR) to proxy Anthropic SDK → Fireworks K2.5
  • Three containers: Temporal + Router (CCR on port 3456) + Worker
  • Portal-only source code uploaded: 55 Go files, 572KB, 15,700 lines
  • Excluded: internal tools, DICOM importers, MCP bridge, patient data, binaries
  • Scan launched: inou-com_shannon-1771049779401 against https://inou.com
  • Fireworks K2.5 cost: ~$0.50 worst case for full scan

Alert Dashboard (port 9202) Redesign

  • Renamed conceptually to "Watchboard" / status board
  • Added DELETE endpoint + × dismiss button on alerts
  • Full redesign: Sora font, Braun/mid-century modern aesthetic, warm gold (#c8b273) accents
  • All 12 clock numbers, gold second hand with counterweight
  • Bigger fonts (16px alerts) for arm's length viewing on Fire tablet
  • Calendar month navigation ( buttons)
  • Compact day headers (S M T W T F S) — fixes Saturday cutoff
  • Two-tone warm chime instead of harsh pling
  • Purpose: status updates (texts, Shannon waiting, sensor alerts), NOT news (news stays on James Dashboard 9200)
  • Johan needs plastic bins for closet: 46W×46D×33H cm space, wants two side by side
  • Each bin ~23cm×46cm×33cm (9"×18"×13") — this form factor doesn't exist in standard retail
  • Suggested: one wider + one narrower bin, or IKEA/Container Store in person

Evening Briefing Test Run

  • Successfully generated and posted 7 news items to dashboard
  • OpenClaw 2026.2.13 release: HuggingFace support, write-ahead delivery queue, Discord voice messages, 337 commits
  • Confirmed briefing format works with individual news items + source URLs

Fully Dashboard Overhaul (2:00am - 2:30am)

Changes Made

  • Alerts moved to left (58/42 split), clock/calendar on right
  • Long-press (300ms) to mark done — dims + strikethrough, auto-purge after 2h
  • Timestamps brightened — was #555, now #777-#888
  • Pulse-ox camera feed — MJPEG stream via HA proxy (/api/cam/pulse-ox/stream), shows 7pm-8am only
    • Entity: camera.pulse_ox_live_view (Tapo camera at 192.168.2.183)
    • HA token passed via systemd env to alert-dashboard service
    • MJPEG stream works natively in <img> tag — no ffmpeg needed
  • Typography cleanup — consolidated to 3 font sizes, 2 weights, consistent spacing
  • Email triage redirected — MC dashboard_url changed from 9202 → 9200 (news API), no more triage noise on Fully board
  • Hook announcements silenced — messages webhook set deliver: false
  • ffmpeg installed on forge (for future use)

Architecture: Fully Dashboard (port 9202)

  • Purpose: Johan's unified inbox / notification center on Fire tablet with Fully Kiosk
  • API: /api/alerts (POST message+priority, GET list, PATCH done, DELETE)
  • SSE: /api/alerts/stream for live push
  • Camera: /api/cam/pulse-ox (snapshot) and /api/cam/pulse-ox/stream (MJPEG)
  • Priority levels: critical (red), warning (amber), info (gold)
  • Sound: Web Audio API chime on new alerts (gain 0.1)
  • Source: /home/johan/dev/alert-dashboard/
  • Service: alert-dashboard.service (systemd user)

Key Decision: Fully Board = Johan's Inbox

  • Everything that survives my triage → Fully dashboard alert
  • Three tiers: critical (act NOW), warning (needs you today), info (FYI)
  • Email, WhatsApp, calendar reminders, system alerts — all go here
  • Johan stops checking email/phone — I'm the filter
  • CMMC meetings: skip by default (Johan doesn't attend)

Calendar Flow (designed, not yet implemented)

  • Johan sends screenshot of Outlook list view each morning
  • I OCR it, parse meetings, set 5min + 1min cron alerts to Fully board
  • List view = machine-readable (exact times, no guessing from pixels)
  • Week of Feb 16-20 parsed as proof of concept

WhatsApp Visibility

  • MC has WhatsApp messages via message-bridge on port 8030
  • Tanya's number: +17272253810
  • She sent laborwerte_1457_tmp.pdf (lab results) at 8:22pm Feb 13
  • Johan wants to be alerted about messages, not have me interpret them

Contacts

  • +13474999485 — sent Johan Valentine's message on WhatsApp (unknown contact, uses "iyawo")

James Dashboard Fix (2:38am)

  • News items now clickable — titles link to source URL when available
  • Briefing agent sometimes grabs wrong URLs (Gemini article linked to Wordle page) — need to improve URL capture in briefing prompts

Johan's Desk Layout

  • Left monitor: TUI console (OpenClaw CLI)
  • Center: Fully dashboard (Fire tablet with Fully Kiosk, port 9202) — promoted to center
  • Center-back: Sophia room camera (Tapo standalone, redundant backup while Fully proves stability)
  • Right monitor: James Dashboard (port 9200)

Sophia Monitoring Stack

  • Primary: VTech radio baby monitor — reliable (radio, not WiFi), beeps on power loss
  • Visual: Tapo WiFi camera (192.168.2.183) + Fully dashboard MJPEG stream
  • Pulse-ox: Visible on Tapo camera feed (typical restful values: 98% SpO2, 73 HR)
  • WiFi camera is nice-to-have, VTech is safety-critical
  • Johan listens to her breathing — prefers hearing it over silence

Email Triage (4:18 PM)

  • Processed 34 messages (19 tj, 15 johan)
  • Generator: 48kw ran 3x today (power outages 11:23, 11:31, 11:48 AM). GenerX tech Eduardo Rivera came for service. Service form archived.
  • Bounces: 10 bounce-backs on johan@ from inou pentest probes (example.com, pentest@inou.com) — expected
  • Deliveries: TP-Link Ultra-Portable delivered. Orders #4720443389 and #12539598 shipped.
  • Archived: GenerX service form, order confirmations (4720443389, Nordstrom 1026793809)
  • Trashed: Duke Energy usage alert, Mint Mobile survey (Mikhail), Realtor listing for own house, generator notifications, political newsletter, subscriber agreement update, marketing/spam, Valentine's lingerie promos, hardware spam
  • MC issue: LLM triage failing with 401 (API key invalid), requests taking 15s+

Shannon Scan Complete

  • Full pentest of inou.com via Shannon Lite on Sonnet 4.5 (OAuth token direct to Anthropic)
  • Runtime: ~1.5 hours
  • Findings: 2 CRITICAL (hardcoded backdoor 250365, session hijacking), 2 HIGH (session fixation, brute force)
  • Authorization, SQLi, SSRF all passed clean
  • 3 out-of-scope items needing internal access (path traversal, DICOM XSS, LLM prompt injection XSS)
  • PowerPoint generated and placed on sophia SMB share
  • All deliverables at ~/clawd/memory/shannon-scan-2026-02-14/

Prima (UMich Brain MRI AI)

  • Deep dive completed — open source MIT, VLM for 50+ neuro conditions, 97.5% accuracy
  • Needs Ampere+ GPU (L40S reference, RTX 3090 minimum). Forge GTX 970 is out
  • Johan's idea: offer as inou service — intelligent series selection (don't run all 10K slices, pick relevant series) → RunPod serverless GPU → results in viewer
  • Spec being written by subagent
  • Sophia had 10,000+ slices — this is the use case

Diana Yusha Lab Report

  • Added arthritis rebuttal section (Russian) — CRP 1.3 and ESR 6 don't support arthritis diagnosis
  • Some Russian doctor suggested arthritis based on elevated CK 406 — incorrect interpretation

WebMCP

  • Added to HEARTBEAT.md tech watch list
  • Chrome experimental API for structured web agent tools — high relevance for James/OpenClaw, medium for inou

Colorado Camel Milk

  • Order #16698 from Feb 7, $115 (6 pints frozen + $31 shipping)
  • Still "Awaiting Shipment" — waiting for safe weather window (frozen perishable to Florida)

Infrastructure

Generator (48kw)

  • Generator does NOT auto-start — manual start only
  • GenerX came out today to analyze the auto-start issue (tech: Eduardo Rivera)
  • The 3 runs were test runs by GenerX, NOT power outages
  • Waiting on quote from GenerX for the fix
  • Installed TP-Link AX1500 router, connected to UDM-Pro port 10
  • Tested and working, then port 10 disabled on UDM-Pro
  • WiFi network: X4 / password: Helder06
  • For longer/full outages (including mobile coverage): swap the cable on port 10 for the longer cable (already staged nearby) and connect to Starlink
  • This gives internet via Starlink → AX1500 → UDM-Pro port 10
  • Shannon VPS now configured with OAuth token (no more router proxy)