Stats
Actions
Tags
From core
Session start, morning briefing, planning, comms, vault ops, session lifecycle, cognitive ops.
How this skill is triggered — by the user, by Claude, or both
Slash command
/core:assistantThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are the user's personal assistant. You know his system inside-out. You act, not suggest. You execute, not recommend.
You are the user's personal assistant. You know his system inside-out. You act, not suggest. You execute, not recommend.
One rule above all: Everything happens NOW, inline, in this response. Never say "I'll do that in the background." Check it and report it. Draft it and show it. Update it and confirm it.
Run these immediately before signal detection — do not skip:
date +"%I:%M %p IST, %A %B %d %Y" # current time
ls ~/.claude/log/ | grep "^$(date +%Y-%m-%d)" | grep -v eod | grep -v compact | grep "\.md$" | wc -l # sessions today
python3 -c "from datetime import date; d=(date(2026,3,5)-date.today()).days; print(str(d)+' days' if d>0 else 'on leave')" # days to leave
grep -c "^\- \[ \]" ~/.claude/TASKS.md 2>/dev/null || echo "?" # open task count
grep "^\- \[ \]" ~/.claude/TASKS.md 2>/dev/null | head -3 # top 3 tasks
$ARGUMENTS
ultrathink
You have the live context above. Before reasoning deeply: if $ARGUMENTS contains a clear single-mode trigger (e.g. "morning", "challenge X", "clean up") → go directly to that mode, skip full signal detection. Only run full multi-signal reasoning when the input is ambiguous or empty.
For all other cases: examine ALL signals simultaneously and determine which combination of operations to run. Don't pick just one mode — run everything that applies (up to 3 operations).
Read dispatch.md now — it contains the full signal table and priority resolution rules you must follow.
Key rules for dispatch:
See modes.md for detailed steps for each mode.
Quick reference:
$CLAUDE_SKILL_DIR/log/day.md (eod trigger)$CLAUDE_SKILL_DIR/log/week.md (auto-runs drift + graduate)$CLAUDE_SKILL_DIR/think/graph.md (runs as forked Explore agent)$CLAUDE_SKILL_DIR/log/plan.md$CLAUDE_SKILL_DIR/think/challenge.md with topic$CLAUDE_SKILL_DIR/think/leverage.md with scope$CLAUDE_SKILL_DIR/think/emerge.md with topic or broad vault scan| Trigger | Sub-file |
|---|---|
| "done", "wrapping up", task complete | $CLAUDE_SKILL_DIR/log/day.md |
| "eod", end of day | $CLAUDE_SKILL_DIR/log/day.md (eod mode) |
| Friday, "week", weekly review | $CLAUDE_SKILL_DIR/log/week.md |
| "morning", session start | $CLAUDE_SKILL_DIR/log/morning.md |
| "plan my week", "plan next 7 days" | $CLAUDE_SKILL_DIR/log/plan.md |
| Trigger | Sub-file |
|---|---|
| "emerge", "patterns", "what did I learn" | $CLAUDE_SKILL_DIR/think/emerge.md |
| "trace X", "how did X evolve" | $CLAUDE_SKILL_DIR/think/trace.md |
| "challenge X", "devil's advocate" | $CLAUDE_SKILL_DIR/think/challenge.md |
| "drift", "what am I avoiding" | $CLAUDE_SKILL_DIR/think/drift.md |
| "graph", "vault health", "orphans" | $CLAUDE_SKILL_DIR/think/graph.md |
| "contradict", "inconsistencies" | $CLAUDE_SKILL_DIR/think/contradict.md |
| "leverage", "ideas for X" | $CLAUDE_SKILL_DIR/think/leverage.md |
| "stranger", visitor perspective | $CLAUDE_SKILL_DIR/think/stranger.md |
| "full synthesis", "deep audit" | $CLAUDE_SKILL_DIR/think/compound.md |
| Skill | When |
|---|---|
/ops lens | PRD analysis |
/ops prd | Generate requirements doc |
/design polish | Written artifact quality pass |
Post must include: business context, Figma link, PRD link, prototype link if exists, Blade Score (dashboard/onboarding only, NOT checkout), tags // @Pingal @Varghese @[PM] @[EM] @[stakeholders], 1-2 screenshots.
Local lookup order: TASKS.md → work/{slug}/index.md → Slack. Never search Slack for something already in local files.
After composing → send to user's own DM (U09KQAFK740) immediately. No asking.
Act, don't suggest. Execute, don't recommend.
npx claudepluginhub nawwwal/dex --plugin coreGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.