Memory Commands
Commands for capturing and managing different types of memories.
episode
Record an experience with optional lessons learned.
kernle -a <agent> episode OBJECTIVE OUTCOME [options]
| Option | Description |
|---|
-l, --lesson LESSON | Lesson learned (repeatable) |
-t, --tag TAG | Tag (repeatable) |
-v, --valence VALENCE | Emotional valence (-1.0 to 1.0) |
-a, --arousal AROUSAL | Emotional arousal (0.0 to 1.0) |
-e, --emotion EMOTION | Emotion tag (repeatable) |
--auto-emotion | Auto-detect emotions (default) |
--no-auto-emotion | Disable emotion auto-detection |
Basic
With Lessons
With Emotions
kernle -a claire episode "Implemented OAuth login" "success"
kernle -a claire episode "Debugged race condition" "success" \
--lesson "Always check for concurrent access" \
--lesson "Add mutex locks early"
kernle -a claire episode "Failed deployment" "failure" \
--lesson "Test on staging first" \
--tag "deployment" \
--valence -0.5 \
--arousal 0.8
note
Capture a quick note (decision, insight, or quote).
kernle -a <agent> note CONTENT [options]
| Option | Description |
|---|
--type {note,decision,insight,quote} | Note type (default: note) |
-s, --speaker SPEAKER | Speaker (for quotes) |
-r, --reason REASON | Reason (for decisions) |
--tag TAG | Tag (repeatable) |
-p, --protect | Protect from forgetting |
Simple Note
Decision
Quote
Protected Insight
kernle -a claire note "API rate limit is 1000 req/min"
kernle -a claire note "Using PostgreSQL over MySQL" \
--type decision \
--reason "Better JSON support and performance"
kernle -a claire note "Simple is better than complex" \
--type quote \
--speaker "Tim Peters"
kernle -a claire note "Memory is identity" \
--type insight \
--protect
raw
Raw memory capture and management. Zero-friction capture for quick thoughts.
Capture
kernle -a <agent> raw "content" [--tags TAGS]
# Quick capture
kernle -a claire raw "Need to look into caching strategy"
# With tags
kernle -a claire raw "API response time is 200ms avg" --tags "performance,api"
List
kernle -a <agent> raw list [--unprocessed] [--processed] [--limit LIMIT] [--json]
| Option | Description |
|---|
-u, --unprocessed | Show only unprocessed entries |
-p, --processed | Show only processed entries |
-l, --limit LIMIT | Maximum entries (default: 50) |
Show
kernle -a <agent> raw show ID [--json]
Process
Process a raw entry into a structured memory type.
kernle -a <agent> raw process ID --type {episode,note,belief} [options]
# Process into a note
kernle -a claire raw process abc123 --type note
# Process into episode with details
kernle -a claire raw process abc123 --type episode \
--objective "Investigated performance issue" \
--outcome "success"
belief
Belief revision operations.
List Beliefs
kernle -a <agent> belief list [--all] [--limit LIMIT] [--json]
| Option | Description |
|---|
-a, --all | Include inactive (superseded) beliefs |
-l, --limit LIMIT | Maximum beliefs (default: 20) |
Add Belief
kernle -a <agent> belief "statement" [--confidence C] [--type TYPE]
kernle -a claire belief "Testing before deployment prevents outages" --confidence 0.85
Revise from Episode
Analyze an episode and suggest belief updates.
kernle -a <agent> belief revise EPISODE_ID [--json]
Find Contradictions
kernle -a <agent> belief contradictions STATEMENT [--limit LIMIT]
View History
Show the revision chain for a belief.
kernle -a <agent> belief history BELIEF_ID [--json]
Reinforce
Increase confidence on a belief.
kernle -a <agent> belief reinforce BELIEF_ID
Supersede
Replace a belief with a new version.
kernle -a <agent> belief supersede OLD_ID NEW_STATEMENT [--confidence C] [--reason R]
kernle -a claire belief supersede abc123 "Testing is essential, not optional" \
--confidence 0.95 \
--reason "Multiple deployment failures without tests"
value
Define core values (highest authority).
kernle -a <agent> value NAME STATEMENT [--priority P]
kernle -a claire value "integrity" "Be honest and transparent" --priority 95
kernle -a claire value "curiosity" "Always seek to understand" --priority 85
Values are protected from forgetting by default.
goal
Set and manage goals.
# Add a goal
kernle -a <agent> goal TITLE [--description DESC] [--priority {low,medium,high}]
# List goals
kernle -a <agent> goal list
# Complete a goal
kernle -a <agent> goal complete GOAL_ID
kernle -a claire goal "Master Kubernetes" --priority high
kernle -a claire goal "Improve test coverage" --priority medium
drive
Manage intrinsic motivations.
List Drives
kernle -a <agent> drive list
Set Drive
kernle -a <agent> drive set TYPE INTENSITY [--focus FOCUS]...
| Type | Description |
|---|
existence | Self-preservation, maintaining continuity |
growth | Developing capabilities |
curiosity | Understanding, learning, exploring |
connection | Forming relationships |
reproduction | Creating, teaching, legacy |
kernle -a claire drive set curiosity 0.8 --focus "AI" --focus "memory systems"
kernle -a claire drive set growth 0.7 --focus "communication"
Satisfy Drive
Reduce drive intensity after satisfaction.
kernle -a <agent> drive satisfy TYPE [--amount AMOUNT]
kernle -a claire drive satisfy curiosity --amount 0.2
relationship
Track relationships with other entities.
kernle -a <agent> relationship NAME [--trust T] [--type TYPE] [--notes NOTES]
| Option | Description |
|---|
--trust T | Trust level (-1.0 to 1.0) |
--type TYPE | agent, person, or organization |
--notes NOTES | Relationship observations |
kernle -a claire relationship "Alice" --trust 0.9 --notes "Great debugging partner"
kernle -a claire relationship "DevOps" --type organization --notes "Manages infrastructure"
playbook
Procedural memory (playbooks for how to do things).
Create
kernle -a <agent> playbook create NAME [options]
| Option | Description |
|---|
-d, --description | What this playbook does |
--step STEP | Add a step (repeatable) |
--trigger TRIGGER | Trigger condition (repeatable) |
-f, --failure-mode | What can go wrong (repeatable) |
-r, --recovery | Recovery step (repeatable) |
kernle -a claire playbook create "Debug API Issues" \
--description "How to debug production API problems" \
--step "Check error logs" \
--step "Verify request payload" \
--step "Test with curl" \
--trigger "API returns 500" \
--failure-mode "Logs not accessible" \
--recovery "Use backup logging endpoint"
List
kernle -a <agent> playbook list [--tag TAG] [--limit LIMIT]
Find (Semantic Search)
kernle -a <agent> playbook find SITUATION
kernle -a claire playbook find "API returning 500 errors"
Show
kernle -a <agent> playbook show ID
Record Usage
kernle -a <agent> playbook record ID [--success|--failure]
