Stats
Actions
Tags
From juggle
Weekly Monday 03:00 routine — queries Juggle DB + Hindsight + auto-memory for a weekly digest (reports/reflect-YYYY-MM-DD.md) committed directly to main, plus up to 5 GitHub issues.
How this skill is triggered — by the user, by Claude, or both
Slash command
/juggle:schedule-reflectThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Goal:** Produce a weekly operational digest covering watchdog health, action item patterns, agent quality, context bloat, memory health, and skill drift. Read-only — no code changes, no edits outside `~/github/juggle/`.
Goal: Produce a weekly operational digest covering watchdog health, action item patterns, agent quality, context bloat, memory health, and skill drift. Read-only — no code changes, no edits outside ~/github/juggle/.
~/github/juggle/reports/reflect-YYYY-MM-DD.md committed directly to mainroutine-reflect) for actionable findings~/github/juggle/cd ~/github/juggle
# Live run (commits digest, files issues)
python3 src/juggle_cli.py schedule-reflect
# Dry run (writes digest to /tmp/schedule-reflect-sample-digest.md, no Git/GitHub)
python3 src/juggle_cli.py schedule-reflect --dry-run
| ID | Source | Section title |
|---|---|---|
| RF-1 | watchdog_events (7d) | Watchdog Health |
| RF-2 | action_items (30d) | Action Item Fatigue |
| RF-3 | agent_completions (7d) | Agent Output Quality |
| RF-4 | messages table (7d) | Context Bloat Candidates |
| RF-5 | Hindsight API (60d+ old) | Memory Health |
| RF-6 | ~/.claude/.../memory/ | Auto-Memory Contradictions |
| RF-7 | skill descriptions vs DB tasks | Skill Drift |
| RF-8 | Most recent dogfood report | Dogfood Pulse |
All sections attempted regardless of failures — partial digest beats no digest.
routine-reflect (auto-created on first use)reflect: <summary> (max 72 chars)reports/reflect-YYYY-MM-DD.md committed to mainroutine-reflect label~/github/juggle/| Failure | Action |
|---|---|
| Hindsight unavailable | Skip RF-5, mark "Hindsight unavailable" in digest |
| auto-memory path not found | Skip RF-6, mark in digest |
| DB query returns 0 rows | Write "No events this week" — positive signal |
| Cost cap ($2.00) | Write partial digest with note, stop additional LLM calls |
| GitHub rate limit | Log remaining issues in digest, retry at next run |
python3 src/juggle_cli.py complete-agent <THREAD_ID> "Reflect complete: reports/reflect-YYYY-MM-DD.md committed. N issues filed. Cost=$X.XX." --retain "Reflect ran YYYY-MM-DD. Key finding: <summary>."
npx claudepluginhub cyc115/juggle --plugin juggleGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.