Sabbatical · self-improvement

Learning curriculum

Catching up on the system you operate — and the capabilities you're not yet using. Extended weekly.

Delivered

10 of 23

Where your automation runs — and where failures go

Sun 5/31 · LaunchAgent vs cron vs /schedule vs claude -p

Reading your own system's state

Mon 6/1 · JARBUS events, the TTR card lifecycle, 'blocked' vs 'error'

Token economics

Tue 6/2 · What's expensive, the 5-min cache TTL, when Opus vs Haiku

Capability Day 4

The Next.js Server/Client boundary

Wed 6/3 · 'use client', why ssr:false is forbidden in Server Components, CSS @import order

deep-research — fan-out and adversarial verify

Thu 6/4 · When to reach for the research harness instead of hand-searching

Capability Day 6

/loop and ScheduleWakeup vs fragile LaunchAgents

Fri 6/5 · Self-paced autonomous work the harness can actually see fail

Capability Day 7

Worktrees and parallel agents

Sat 6/6 · Testing 3 versions of an idea cheaply, in isolation

Capability Day 8

Memory hygiene — the absence-claim trap

Sun 6/7 · Why 'X doesn't exist' poisoned a session (#1776) and how to correct it

Operating the machine vs building it

Mon 6/8 · Putting it together — steering the portfolio, not just the pipeline

Agent view — running many sessions at once

Tue 6/9 · Claude Code's new agent view, --bg background sessions, /resume, peeking at status

Capability Day 11

Supabase — the database under Focaccia Vote

Wed 6/10 · What Postgres + Supabase are, tables vs rows, the votes table you filled last week

Row-Level Security — why your data can't leak

Thu 6/11 · RLS policies, the rule that gated the Focaccia votes table, anon vs service keys

Cloudflare Pages — how zorc.app goes live

Fri 6/12 · Pages vs Workers, wrangler deploy, _redirects, custom domains (the 404 you fixed)

Dynamic Workflows — orchestrating many agents

Sat 6/13 · The Workflow tool, fan-out vs pipeline, when it beats a single agent

Capability Day 15

DESIGN.md — why every project carries its own design system

Sun 6/14 · Per-project DESIGN.md, the Instrument Panel redesign, design tokens, /frontend-design

Keeping the portfolio honest

Mon 6/15 · Why Runway Model stalled 13 days unnoticed, STATUS drift, the shepherd daemon

LaunchAgents & launchd — how your cron jobs actually run

Tue 6/16 · launchd, plists, why two of your jobs silently unloaded (the Jun 5 notification audit)

Headless Claude — `claude -p` and the fallbackModel safety net

Wed 6/17 · How plan-day runs unattended, what --fallback-model does, the new June config that retries on overload

Capability Day 19

Watchdogs & timeouts — why a 30-min cap killed a 34-min job

Thu 6/18 · Wall-clock watchdogs, gtimeout --kill-after, Automation Hygiene rule 2 (the May 30 plan-day overrun)

Retry & backoff — the Plaud ingest resilience pattern

Fri 6/19 · The three-tier resilience you shipped this week: in-run retry, stuck-recording giveup, single-alert threshold

Capability Day 21

Idempotency — why a pipeline must give the same result on re-run

Sat 6/20 · The render regression where a second run left briefings/task-boost/languages empty (#3354)

The standards-checker gate — how bad reviews get blocked before they reach you

Sun 6/21 · The quality gate in report_quick_deliver, blocker verdicts, the shared review-standards rubric

reporter.py — the one door JARBUS uses to write to PDB

Mon 6/22 · The load-bearing boundary: why JARBUS never writes PDB state directly, the four report_* functions

Authored daily · surfaced in /plan-day · emailed 10:30am