Instinct Outcome Command

Record an outcome for a specific instinct. This updates its confidence score and outcome counters.

Usage

/instinct-outcome <id> <good|bad|neutral>
/instinct-outcome <id> good "reason why it worked"
/instinct-outcome <id> bad "reason it caused problems"
/instinct-outcome list

Implementation

Run the outcome tracker script:

node "${CLAUDE_PLUGIN_ROOT}/scripts/instinct-outcome-tracker.js" <id> <outcome> [reason]

Or if CLAUDE_PLUGIN_ROOT is not set (manual installation):

node ~/.claude/scripts/instinct-outcome-tracker.js <id> <outcome> [reason]

To list available instinct IDs:

node "${CLAUDE_PLUGIN_ROOT}/scripts/instinct-outcome-tracker.js" --list

Confidence Adjustments

Outcome	Confidence change
good	+0.05 (max 0.95)
bad	-0.10 (min 0.10)
neutral	no change

Preconditions

Before running, verify:

• ~/.claude/homunculus/ exists — if not, instruct user to run install.sh --enable-learning
• The instinct ID is valid — run list subcommand if unsure of the ID
• Outcome is one of good, bad, or neutral

Execution

1. Ask the user which instinct they want to record an outcome for (or use the list subcommand)
2. Validate the outcome is good, bad, or neutral
3. Run the tracker script with the instinct ID and outcome
4. Show the updated confidence score
5. Suggest running /instinct-report to see the updated ranking

Example Output

Recorded good outcome for "test-first-workflow"
  Confidence: 0.70 ↑ 0.75
  Reason: TDD caught a regression before it hit staging

Output Interpretation

• Confidence increased: instinct is gaining evidence — will appear higher in /instinct-report
• Confidence decreased: if drops below 20%, it will be flagged in /instinct-report for removal review
• Run /instinct-report after recording: to see the updated ranked list with trend indicators (↑↓→)
• Pattern: if the same instinct consistently gets bad outcomes, consider using /evolve to revise it

Notes

• Outcomes are logged to ~/.claude/homunculus/outcomes.jsonl for audit
• last_used is updated to today on every outcome recording
• usage_count increments on every outcome recording
• Use /instinct-status to see current confidence levels
• Use /instinct-report to see ranked list with trends

After This

• /evolve — trigger instinct evolution when enough outcomes are collected
• /instinct-report — see the updated ranked instinct list with trend indicators
• /instinct-status — check current confidence levels across all instincts