Back to skills
SKILL.md
008 Operational 04e82941
ASecurityOperational cadence for running LeanOS Core. Covers daily, weekly, and trigger-based routines. ---
- 9 stars
- 0 votes
- 0 copies
- 1 view
- Added October 11, 2026
Security analysis
100/100npx -y skills add tools-only/X-Skills --skill 008-operational_04e82941 --agent claude-codeAre you the author of 008 Operational 04e82941?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/tools-only-008-operational-04e82941)# Daily Routine
Operational cadence for running LeanOS Core. Covers daily, weekly, and trigger-based routines.
---
## Quick Reference
| Time | Routine | Key Actions |
|------|---------|-------------|
| Morning | Start of Day | Review goals, check threads, set priorities |
| Throughout | Work | Progress threads through causal flow |
| End of Day | Close of Day | Update threads, capture learnings |
| Weekly | Review | Goal tracking, thread cleanup, planning |
---
## Morning Routine
### 1. Review Active Goals
```
SCAN strategy/goals/active/*.md
FOR EACH goal with status = activated:
+-- Check: thread progress
+-- Check: blockers
+-- Check: today's milestone
```
### 2. Check Active Threads
```
SCAN threads/*/
FOR EACH thread with open actions:
+-- Read 5-actions.md
+-- Identify: what's next?
PRIORITIZE by:
1. Critical goal threads
2. Time-sensitive threads
3. Quick wins
4. Deep work threads
```
### 3. Set Today's Focus
Pick your top 3 goal-linked tasks for the day.
---
## Throughout the Day
### Thread Progression
```
FOR active thread:
1. READ current state
+-- Where are we in causal flow?
2. EXECUTE next stage
+-- 2-hypothesis.md --> Form hypothesis (use rsn-reasoning-problems)
+-- 3-implication.md --> Derive implications
+-- 4-decision.md --> Make decision (use rsn-reasoning-problems.dialectical)
+-- 5-actions.md --> Execute actions
3. UPDATE thread
+-- Document progress in 5-actions.md
4. CHECK completion
+-- IF actions complete:
+-- Write 6-learning.md (use rsn-learning-outcomes.reflection)
+-- sys-tracking-goals updates goal
```
### Quick Decisions
```
FOR items < 5 min:
- Handle immediately
- Don't create thread (overhead > value)
FOR items > 15 min:
- Create thread
- Link to goal
```
---
## End of Day Routine
### 1. Update Active Threads
```
FOR EACH thread worked today:
+-- Update 5-actions.md with progress
+-- Note blockers
+-- Identify next action
```
### 2. Capture Learnings
```
IF significant learning today:
+-- Write to relevant thread's 6-learning.md
LEARNING TRIGGERS:
- Something failed unexpectedly
- Something succeeded unexpectedly
- Pattern noticed
- Process improvement idea
```
### 3. Check Goal Progress
```
FOR goals worked today:
+-- Update current_value if measurable
+-- sys-tracking-goals will recalculate gaps
```
---
## Weekly Routine
### 1. Goal Review
```
INVOKE sys-tracking-goals (full scan)
FOR EACH active goal:
+-- Review: target vs current
+-- Review: trajectory (on pace?)
IDENTIFY:
- Goals on track (continue)
- Goals at risk (intervention needed)
- Goals achieved (close)
- Goals to abandon (document why)
```
### 2. Thread Cleanup
```
FOR stale threads (no activity > 7 days):
+-- Decide: continue, pause, or close
+-- IF close: write 6-learning.md
FOR completed threads:
+-- Ensure 6-learning.md complete
```
### 3. Learning Aggregation
```
COLLECT learnings from:
- Thread 6-learning.md files
- Failed experiments
- Successful tactics
IDENTIFY patterns:
- What's working?
- What's not?
```
### 4. Next Week Planning
```
FOR EACH goal:
+-- What must happen next week?
+-- What threads need activation?
SET weekly priorities (3 max)
```
---
## Trigger-Based Routines
### On New Opportunity
```
1. INVOKE sys-defining-goals
2. INVOKE sys-activating-goals
3. Begin causal flow in thread
```
### On Experiment Completion
```
1. READ results
2. INVOKE rsn-reasoning-problems.inductive (what patterns?)
3. DECIDE: continue / pivot / stop
4. UPDATE 6-learning.md
```
### On Planning Cycle
```
1. REVIEW canvas (still accurate?)
2. SET new goals (sys-defining-goals)
3. DECOMPOSE if needed (sys-decomposing-goals)
4. ACTIVATE (sys-activating-goals)
```
### On Pivot Decision
```
1. DOCUMENT why (6-learning.md)
2. ASSESS canvas impact
3. RE-RUN fnd-architect + fnd-researcher
4. RE-DEFINE goals from updated canvas
```
---
## Solo Operator Tips
**Batch similar work.** Process all thread updates together. Reduces context-switching.
**Use threads as memory.** Write to thread immediately. Trust the system to remind you.
**Weekly review is non-negotiable.** Skip daily routine sometimes. Never skip weekly review.
**Limit active goals.** 3-5 max. Fewer goals, more progress.
**Capture learnings immediately.** Learning fades fast. Even one sentence helps.
---
## Quick Commands
```bash
# Morning
"What goals need attention today?"
"What threads are active?"
# Thread work
"Continue thread for [context]"
"Update actions for [thread]"
"Complete thread and capture learnings"
# Reasoning
"Figure out why [problem]"
"Brainstorm options for [challenge]"
"What patterns do you see in [data]?"
# Goal tracking
"How are we tracking against [goal]?"
"Run full goal review"
# Foundations
"Help me define my business mode"
"Size my market for [idea]"
"Analyze competition in [space]"
```
Attribution
Comments
Loading comments…
Newsletter
Get the weekly list of safe new skills, and the ones we pulled.