Skip to content

Cascadia OS 2026.6 — Dual Email Pool & Reliability

Latest
Latest

Choose a tag to compare

View all tags
@zyrconlabs zyrconlabs released this 17 Jun 04:01
2026.6
59532b5

Cascadia OS 2026.6 — June 2026

The outreach and reliability release.

Legal Notice: This software is provided for legitimate email business communication.
Users are responsible for compliance with CAN-SPAM, CASL, GDPR, and other applicable
laws in their jurisdiction.

Email & Outreach

Dual sender pool replacing the single-account setup. email-01 handles up to 100 emails per day and fills first. email-02 (separate SMTP infrastructure) handles up to 50 per day and activates automatically when email-01 hits its daily cap. Total daily capacity increased to 150 across independent infrastructure. Both inboxes are polled every 5 minutes with replies tagged by receiving account. Follow-ups are locked to the original sender account and held rather than rerouted if that account is at its daily cap.

PRISM dashboard shows live slot cards for each account — sent today, daily cap, progress bar, and remaining count — updated every 30 seconds.

RECON Lead Engine

Fixed a critical LLM context overflow bug where large search result payloads caused the model to emit truncated JSON and discard every candidate. Also fixed a category rotation deadlock where a failing LLM prevented category cycling. Added interleaved query selection: CoverageManager runs on even cycles, task.md suburb queries run on odd cycles, giving parallel suburb coverage without waiting for ring exhaustion. Search timeout reduced to prevent hanging on dead connections.

Scout Lead Capture

Added a 9-question guided lead capture flow across 4 stages: qualify (business type, team size, pain point, decision maker), scope (volume, timeline), contact (name + email or phone — mandatory), and close (booking). The booking link is blocked until contact information is collected. Owner is notified immediately when a warm lead is ready regardless of contact status.

System Reliability

  • Health watchdog monitors 11 services every 5 minutes with Telegram alerts on state change
  • CRM auto-sleep via OperatorManager (lifecycle: on_demand) — sleeps automatically after 15 minutes of inactivity, wakes on demand
  • RAM pressure governor — background tier operators sleep automatically at memory thresholds
  • Scout session TTL cleans up idle sessions every 30 minutes
  • CRM SQLite vacuum runs weekly

Telegram Commands

  • /email_status — today, 7-day, 30-day breakdown with per-account counts (email-01 / email-02), outreach vs follow-up split, live pool caps, lead pipeline counts
  • /crm sleep and /crm wake — one-tap CRM lifecycle management
  • /version — running version, operator count, node info
  • /outreach N — queue exactly N drafts for approval (1–50, requires human approval before any send)

Fixes

  • Sender account, message type, and slot label now recorded on every send
  • CRM outreach tracking fields unblocked — written after every successful send
  • Slot labels standardized to email-01 / email-02 across all surfaces
  • Backup and temporary files removed from git tracking
  • config.json excluded from version control

Changed

  • RECON task updated: broader city coverage, higher lead target
  • CRM process lifecycle transferred to OperatorManager exclusively
  • RAM governor: PRISM, CRM, QUOTE, and Pulse in background tier
  • PRISM version updated to v2026.6
  • /outreach N cap raised to 50 — user selects exact batch size per run

Full changelog: CHANGELOG.md

License: Apache 2.0 — open-core platform. Commercial operators are maintained separately.

Assets 2
Loading